Reforming to Survive:The Bolshevik Origins of Social Policies

Abstract

We detail how elites provide policy concessions when they face credible threats of revolution. Specifically, we discuss how the Bolshevik Revolution of 1917 and the subsequent formation of Comintern enhanced elites’ perceptions of revolutionary threat by affecting the capacity and motivation of labor movements as well as the nature and elites’ interpretation of information signals. These developments incentivized elites to provide policy concessions to urban workers, notably reduced working hours and expanded social transfer programs. We assess our argument by using original qualitative and quantitative data. First, we document changes in perceptions of revolutionary threat and strategic policy concessions in early inter-war Norway by using archival resources. Second, we code party-representatives at the 1919 Comintern meeting to proxy for credibility of domestic revolutionary threat in cross-national analysis. States facing such threats expanded various social policies to a greater extent than other countries, and some of these differences persisted for decades.

Full text

Reforming to Survive:
The Bolshevik Origins of Social Policies
Magnus Bergli Rasmussen
Carl Henrik Knutsen
University of Oslo
Words: 11981
We detail how elites provide policy concessions when they face credible threats of revolution.
Specifically, we discuss how the Bolshevik Revolution of 1917 and the subsequent formation of
Comintern enhanced elites’ perceptions of revolutionary threat by affecting the capacity and
motivation of labor movements as well as the elites’ interpretation of information signals. These
developments incentivized elites to provide policy concessions to urban workers, notably reduced
working hours and expanded social transfer programs. We assess our argument by using original
qualitative and quantitative data. First, we document changes in perceptions of revolutionary threat
and strategic policy concessions in early inter-war Norway by using archival resources. Second, we
code, e.g., representatives at the 1919 Comintern meeting to proxy for credibility of domestic
revolutionary threat in cross-national analysis. States facing greater threats expanded various social
policies to a larger extent than other countries, and some of these differences persisted for
decades.
* We are grateful for comments and advice from Aasmund Egge, Herbert Obinger, Nik Brandal, Nils Ivar Agøy, Pablo Balan,
Anders Sundell, Andrej Kokkonen, Haakon Gjerløw, Lucas Leemann, André Walter, Damian Bo, Klaus Petersen, Tore Wig, Øyvind
S. Skorge, Kalle Moene, Henning Finseraas, Andreas Kotsadam, Axel West-Pedersen, Jonas Pontusson, Sirianne Dahlum, Jan
Torell, Alice Evans, and Pseudoerasmus. Several participants at EPSA and APSA 2019 gave valuable and constructive criticisms.
Part of this paper was written on free espresso supplied by the Fuglen coffee rostary (Oslo). We are very grateful for archival
assistance from Jorunn Pedersen at ARBARK and research assistance from Solveig Bjørkholt.

1
1. Introduction
The Bolshevik revolution in Russia in 1917 marks the beginning of what Hobsbawm (1994) labels
the “short 20th century”, characterized by the world being divided into a capitalist- and communist
sphere. Yet, the Bolshevik revolution -- a worker-led revolution with aspirations far beyond Russia
-- also constituted a threat to elites in capitalist countries. The Bolshevik revolution was a symbolic
event and learning experience for revolutionaries globally (Pons 2014). The Bolsheviks also
promised ideological and logistic support to revolutionaries outside Russia. This contributed to
split many labor movements between reformist and revolutionary groups, with the formation of
several new communist parties (Lipset 1983; Pons 2014). We propose that these developments,
and the resulting new threats to the political and economic orders cherished by elites in different
countries, helped shape policy-making even in countries that remained staunchly capitalist.
Previous theoretical and empirical work (Acemoglu & Robinson 2006; Aidt & Jensen 2014;
Przeworski 2009) have highlighted how revolutionary threats by the poor may spur franchise
extension. Other studies have documented that business elites’ preferences for accepting
redistributive policy is contingent on perceiving revolutionary threats (Paster 2013). This could
contribute to explain the finding that mass mobilization and destruction during and following the
world wars led to e massive expansions in social spending (Obinger & Petersen (2017). Previous
work has also documented how social policy can be used to enhance regime legitimacy (e.g.
Rimlinger 1971). There can also be international dimensions to such dynamics; notably, the
competition between “East” and “West” during the cold war increased political incentives for
enacting generous social policies in both camps (Obinger & Schmitt 2011).
Integrating these different insights, we theorize and assess how revolutionary threats may
shape politics, and we consider policy change rather than regime change. Further, we move the
focus from long-term legitimacy concerns and rather consider how social policy can be used to
coopt immediate threats. We also go beyond business elites and focus on how various political

2
elites respond to revolutionary threats, specifying the cognitive and rational mechanisms that
induce elites to expand social policy to mitigate perceived threats.
We embed this general argument in a particular context. Specifically, we address how
developments following the Bolshevik Revolution – and especially after the formation of
Comintern in 1919 – shaped social policies by spurring perceptions of credible revolutionary
threats among elites, who responded with expanding social policies aimed at defusing the labor
threat. Comintern participation by domestic worker organizations was arguably related to enhanced
(perceptions of) working class revolutionary threats, potentially due to multiple mechanisms: First,
the international network and Russian support that accompanied Comintern participation
increased the capacity of domestic revolutionary actors. Second, participation in Comintern altered
these groups’ ideological outlook, if nothing else by strengthening more radical and revolutionary
fractions. Third, working class organizations being invited by the Russians to attend Comintern
functioned as an easy-to-identify informational cue that enhanced elite perceptions of a credible
revolutionary threat.
The exact relevance of these channels notwithstanding, an increase in perceived
revolutionary threat, here following the Bolshevik revolution and subsequent formation of
Comintern, may have induced elites to pursue large-scale expansions of social transfer programs,
reduce working hours, etc., to appease the working classes and quell grievances that could fuel a
revolution. We highlight that we do not aim to explain all variation in our outcomes of interest.
Our focus on revolutionary threats does not imply that other proposed determinants of
redistributive policies and social transfer programs, are irrelevant. Some of these factors, such as
suffrage extensions, may even be potential transmitters of (parts of) the relevant effect from
revolutionary threat to policy change, and yet other factors, such as war mobilization and cross-
border policy diffusion, may be orthogonal. To the extent possible, our empirical analysis also tries
to ensure that our relationship of interest is robust to accounting for relevant, alternative
explanations.

3
Our empirical analysis is two-fold, comprising an in-depth historical case study of Norway
1915–1924 and cross-country analysis using different new proxies of revolutionary threats and
various new measures of policy characteristics. These analyses provide complementary pieces of
evidence that support our argument that fear of revolution drove elites to extend various social
policies, as concessions, after the Bolshevik revolution and formation of Comintern.
For our case study, we draw on numerous archival and other Norwegian-language sources
to document perceptions on likelihood of revolution from employer organizations, police, military
officers, and politicians, and their strategies for countering this threat. Major Norwegian welfare
expansions, at least from the 1930s onwards, have typically been interpreted as resulting from social
democratic reformism (Esping-Andersen and Korpi 1986) or cross-party welfare state consensus
(Kuhnle 1981). Norway may therefore be considered a fairly tough test case for the proposition
that revolutionary fear pushed elites to adopt social policy as conciliatory measures. More
importantly, pressures for social policy adoption were not directly tied to mobilization during World
War I, as Norway remained neutral. Hence, this case selection removes one plausible source of
confounding. Instead, we detail how Norwegian elites believed that workers could conduct a
revolution following the Bolshevik revolution, and institutional linkages between the Labor Party
and Comintern helped shape this belief. Economic and political elites coordinated their response
by using various stick- and carrot-tactics. Regarding the carrots, elites strategically pursued
appeasement and inclusion, implementing several reforms that they initially had opposed, especially
the eight-hour workday. Following expectations, gains granted before the revolutionary fear
dissipated (in 1923/24) proved persistent, whereas promised policies not yet passed were never
implemented.
For our cross-country tests, we code and use new country-level measures on social policies
and on labor regulation. Further, we code new proxies of revolutionary threat (perceptions), for
example coding the formation of soldier and worker councils. Yet, our main measure draws on the
fact that Trotsky in January 1919 invited several revolutionary groups to set up Comintern in

4
Moscow (Carr 1979). Invitations did not include all labor organizations and were not random, but
only sent to truly radical worker groups (to avoid “ideological contamination”). This allows us to
distinguish contexts where labor had adopted a radical ideology from others. Not only should we
consider these Russian invitations a true “expert opinion” on which countries faced revolutionary
pressures in 1918-19, they also provided clear signals to elites that domestic labor groups were
revolutionary and had significant resources at their disposal. Countries facing such a clear threat
were more likely to limit working hours and pass more extensive and generous social policy
reforms. Various panel regressions using alternative control strategies, instrumental variable
analysis, and synthetic control analyses – most of which are included in the online appendix to
focus the presentation – corroborate these results. This revolutionary shock lingered on; states that
experienced greater revolutionary threat in 1919 had lower working hours at the end of the Cold
War. Further analysis (Appendix A13) suggests that this persistent relationship, in part, relates to
the historical formation of Communist parties, linked to Comintern and funding from Moscow.
Below, we first present the historical research context, before we present our more general
theoretical argument on which features enhance revolutionary threats, as perceived by incumbent
elites. In extension, we discuss how perceived threats may spur social policy legislation. In the third
and fourth sections, we present the historical case study of Norway and cross-country analysis,
respectively. Finally, we provide a summary and detail how our study contributes to several research
literatures.
2. Argument
2.1 The Bolshevik Revolution and formation of Comintern
The Russian Revolution of 1917 would eventually bring the Bolsheviks to power, introducing a
Communist state that would shake up the international system. The Bolsheviks saw their revolution
as a prelude to World Revolution (Pons 2014, 15), and even considered revolutions abroad as key
to their own long-term (regime) survival (Pons 2014, 8-9; Car 1979, 12-13). Comintern was thus
established to guide revolutionary groups outside of Russia (Agnew & McDermott 1998).

5
Comintern was de facto controlled by the Russian Communist Party’s Politburo, and worked
alongside the Russian Ministry of Foreign Affairs (Pons 2013, 12). Its mission, in Trotsky’s (1919)
words, was to bring together the world’s true revolutionaries, eschewing the “reformist class
traitors”. It established local branches to spread propaganda, and later required that all member
parties adhere to Lenin’s 21 points from the second conference, including that capitalist societies
had progressed to a state of “civil war” (Sundvall 2017). Comintern also provided extensive funding
for aspiring communist parties (Agnew & McDermott 1998).
Carr (1979, 13) notes that, for a while, “the hope of a world revolution seemed too
materialize”. Worker- and soldier councils, strikes, and uprisings surged in many countries. Still,
hopes of worldwide revolution were short-lived; no durable communist regime outside Russia
materialized. In 1921, Lenin admitted that the revolutionary trajectory had “not been as linear as
we had expected”. Why was the Bolshevik revolution not followed by similar successful revolutions
elsewhere? But, even if it did not result in successful revolutions, did it lead to major political and
policy changes outside Russia? One resolution to answering both questions lies in recognizing that
while the revolution was a symbol for labor and revolutionaries globally, it also inspired counter-
strategies by economic and political elites. Fearing revolution, elites responded with several political
and economic reforms, aiming to appease and create vested interests among (parts of) the labor
movements. Below, we outline a more general argument on why and how elites respond to credible
revolutionary threats by strategically providing policy concessions.
2.2 Elites, preferences and coalitions
Why would economic and political elites accept extensive social policy arrangements that (often)
redistribute resources to relatively poor citizens?
1
The welfare state literature and theories of
1
By elites we mean groups of wealthy or powerful individuals with intrinsic interests in maintaining
the current economic system (e.g., capitalism with private property) and existing political regime.
In our context, elites encompass established conservative- and liberal party leaderships, but also

6
democratization provide different responses; at the risk of oversimplifying, we distinguish between
cross-class theories and class-based theories. The former argue that certain anti-elite groups and
some elites can form pro-redistribution, cross-class coalitions, against other elites. The latter
assumes that redistribution is a zero-sum game between elites and non-elites, and a key factor
behind redistributive policies is thus the institutional environment and how it affects the power
balance between elites and other citizens.
One strand of cross-class theories focus on business interests, arguing that employers and
their political representatives can gain from welfare state development and labor regulation (Iversen
& Soskice 2019; Martin & Swank 2012). Sectoral features related to exposure to international
competition, or corporatist organizational structures could drive some employers to support
welfare policies (Martin & Swank 2012). Despite these incentives for certain elite groups, power
resource theories and class theories of democratization (e.g., Esping-Andersen & Korpi 1986; Boix
2003; Acemoglu & Robinson 2006; Korpi 2006) suggest that elites are often inherently resistant to
regulations or redistributive policies. Yet, even if elites are inherently resistant, our theoretical
argument below details how, and in which contexts, elites may change from being antagonists to
consenters to expanding social policy.
2.3 Power resources, ideology and signals
We argue that elites become consenters to social policies when they face a credible revolutionary
threat to the political and/or economic system. Elites would rather consent to social policies and
lose out monetarily from redistribution in its milder form (e.g., increased taxation and spending on
social programs) than risk revolution and more extreme changes in power structure and
redistribution (nationalization, collectivization, etc.).
employers, capitalists, and landowners (and their organizations). Hence, we do not assume that
elites are unitary actors, but we do assume that different elites prefer to maintain the status quo
over revolution and may try to coordinate their policy-responses to revolutionary threats.

7
This argument follows key rationalist contributions to the democratization literature (e.g.,
Aidt & Jensen 2014; Acemoglu & Robinson 2006; Boix 2003) in highlighting that elites may provide
concessions to the lower classes to avoid revolution. In these works, concessions come in the form
of political liberalization and suffrage expansions, which shifts political power to the lower classes,
thereby ensuring future redistribution. Yet, and counter to Acemoglu & Robinson’s (2006), we
highlight that regime change and franchise expansion are not necessary requisites for credible
guarantees of future redistribution (though they certainly help). Instead, we follow Knutsen and
Rasmussen (2018), who highlight that implementing new social programs, and other major policy
initiatives, create lock-in effects and serve to tie elites to these programs once initiated. Not only
do sunk costs with starting up programs (investment in monitoring capacity, complementary
human capital investments, infrastructure, etc.) create lock-in effects once investments are made,
but discontinuing popular programs also create clear focal points on which opponents of the
regime/policy change may organize effective, large-scale opposition. Therefore, social policy
programs, once implemented, generate fairly credible guarantees for future redistribution to its
recipients. In contrast, mere promises of future policy initiatives lack these features, and are
therefore often non-credible (also suggested by our case-study).
If credible revolutionary threats drive elites to provide social policy concessions, a decisive
question for understanding welfare state expansion is what factors shape elites’ perceptions of
likelihood of revolution? We build on Paster’s (2013) work on employers’ sensitivity to
revolutionary threat, but further detail the rational and cognitive dimensions of threat perceptions
of (also political) elites. Below we outline the factors that could shape both the objective probability
of successful revolution and elites’ estimates of this probability.
First, opposition groups must possess power resources to be a viable threat. For workers,
key resources have historically resided in hierarchical and effective organizations that enable them
to solve collective action problems and mobilize large numbers, including trade unions, councils,
and party organizations. International linkages is another key factor. Such linkages may provide

8
monetary resources and organizational know-how, even for smaller groups. Notably, our empirical
study highlights how Comintern in advanced domestic organizational structures, including new
Communist parties.
Second, for revolutionary threats to be credible, opposition groups should espouse an
ideology of radical societal transformation, for instance aiming to transform the economy’s
ownership structure by socializing property. Pursuing legislative change through parliament may
also explicitly be rejected, with movements instead spurring extra-parliamentary action such as
strikes or mass-protests, even considering violent means as legitimate. Historically, this description
fits to parties and unions adhering to Communism or the Zimmerwald movement, as opposed to
the reformism of social democrats (Lipset 1983).
Third, elites must receive some sort of informative signal, i.e., some indication of the
intention and resources of opposition groups, which they must subsequently interpret (Fearon
1994; Vis 2019; Weyland 2019). The nature and interpretation of such signals shape elites’
perceptions of credibility of revolutionary threats. Sometimes revolutionary movements may send
strong and clear signals on their motivations and resources. In other contexts, such signaling is
difficult. Certain reformist labor unions or parties unwilling to engage in revolutionary activities
might have incentives to “bluff”, and pose as revolutionary, to obtain concessions. Thus, elites may
have a hard time distinguishing revolutionaries from reformists.
Elites’ capacity to absorb signals, and how they go about in interpreting them, are also
relevant factors. While some elite actors may decipher information about the motivation and
capacity of opposition groups, and update beliefs in a relatively unbiased and “rational” manner,
others make decision under uncertainty and time-pressure, resorting to use various cognitive short-
cuts (Vis 2019; for applications to democratization, see Weyland 2019). Revolutions are complex
and relatively rare events, making it even harder for elite actors to analyze prospects of revolutions
without relying on cognitive heuristics. Bluntly assuming that the future will reflect the past, elites
may thus evaluate ongoing processes by searching for patterns and analogous events found in past

9
revolutionary settings (Aidt & Jensen 2014), for which the Bolshevik Revolution long remained a
primary reference point (Hobsbawm 1994). Moreover, given the so-called “availability heuristic”
(Vis 2019), elites, as other people, may inadvertently focus on large and salient events, thus over-
estimating the baseline probability of revolution. Since big historical events create symbols that
work as cognitive maps to understand current events, the Bolshevik revolution is likely to have
formed many elites’ perceptions of domestic conditions.
In our research context, ties with the Russian revolutionary regime, especially Comintern
membership from 1919, may have enhanced all the three above-discussed aspects. First, Comintern
often provided material resources directly to relevant movements and helped the founding of new
Communist parties, globally, thus enhancing organizational capacity. Second, the related
international network and exchanges presumably diffused revolutionary ideology. Third,
Comintern membership served as a strong signal to elites, increasing perceptions of a domestic
revolution being likely.
2.4 Elite responses
When facing (perceived) revolutionary threats, elites may respond by pursuing co-optation
strategies. Co-optation can come in at least two forms: political co-optation includes granting
participation rights to previously excluded groups. Economic co-optation includes granting greater
access to material resources. Both strategies aim at defusing revolutionary threats by increasing the
legitimacy of the current system and mitigating core grievances within key opposition segments via
inclusion. We focus on economic inclusion in the form of social policies. Political inclusion is
beyond the scope of this paper. Yet, we do expect such alternative inclusion strategies are also
pursued by elites under conditions of high revolutionary threat – as highlighted by Acemoglu &
Robinson (2006), providing political rights is a key strategy for mitigating revolutionary threats (in
addition to directly providing policy concessions, as we discuss below). Down the road, such
expansions of political rights may also induce politicians to adopt more expansive social policy

10
programs to cater to the recently included groups, thus suggesting one additional long-term
mechanism through which revolutionary threat may affect our outcomes of interest.
2
However, we discussed how elites may credibly respond to revolutionary threats also when
stopping short of extending political rights, instead co-opting threats by implementing reforms and
changing various policies in directions desired by the potential revolutionaries. The group that
constituted the primary threat to elites – especially in industrializing or industrialized countries – in
the period that we study was urban labor, often organized in labor unions, councils, and socialist
parties (Ruschmeyer et. al. 1992). Hence, policies introduced to stem revolutionary threats should
primarily target this group. The important demand made by close to all labor organizations was the
eight-hour day/48-hour week -- made already by the First Socialist International in the 1860s, but
elites in various countries had resisted this policy change. It was one of the first key changes made
by the Bolsheviks in November 1917, setting a concrete standard for labor movements where
revolution had not (yet) taken hold. If employers and governments wanted to co-opt labor to avoid
revolution, the eight-hour day would be a prime tool for doing so.
3. Case study: The Bolshevik Revolution and revolutionary fear in Norway
In Appendix B, we present a detailed case study of early-20th-century Norway, drawing on
numerous archival and other sources. Here, we present the condensed version, discussing how
various Norwegian elites thought that revolution was imminent, between 1918 and 1921. We
2
Indeed, our quantitative measures of revolutionary threat correlate positively with subsequent
changes in our measures of electoral democracy and suffrage extension (from V-Dem; Coppedge
et al. 2018) in regression specifications similar to those in Table 3. This result corroborate those
presented in Aidt and Jensen (2014) and Przeworski (2009). However, when we control for suffrage
in our benchmark specifications on different policy outcomes (Appendix A5), our measures of
revolutionary threat remain robust. This suggests that revolutionary threat drives policy change even
when we hold impulses via suffrage extensions constant.

11
further show how these elites, in response, pursued a complex “sticks and carrots” strategy to avoid
a revolution. Regarding the carrots, we focus on the eight-hour day. However, the trajectory and
timing of policy-making is similar for several other policies. In the appendix, we describe
developments for co-determination, changes in electoral rules, collective agreements, and old-age
pensions. The elite’s aim of pursuing these measures was to incorporate and strengthen the
reformist part of the labor movement (Fure 1983; Knutsen 1994, 43-46).
3.1 Norwegian labor going revolutionary
The Norwegian Labor Party (Det Norske Arbeiderparti [DnA]), founded in 1887, was inspired by
Marxist thinking from its inception. The first party program stated that DnA «endeavors the
handing over of the means of production to social common property and production changed
from capitalistic to socialistic” (DnA 1891). Such ideas were slowly abandoned as reformism
became the leading principle before the turn of century, acknowledging the importance of
achieving social change through legislation in parliament. The trade union federation (Arbeidernes
Faglige Landsorganisasjon [AFL]), founded at DnA’s behest in 1899, was a means to achieve
changes in wage- and working-conditions through bargaining with employers. DnA and AFL were
bound by the hip, and dual membership was practiced. The unions and party coordinated their
demands against employers and politicians.
In early 1917, reformist leaders such as Olaf Lian held leadership positions in both DnA
and AFL. Radical elements were delegated to a minority position, largely originating from
“Fagoposisjonen” of 1911 – a syndicalist movement within AFL – and the social democratic youth
organization. Fagoposisjonen argued for sabotage, boycott, and sympathy strikes as legitimate
weapons against employers (Olstad 1998, 173). Its leader, Martin Tranmæl, wanted to radicalize
the union movement, “to make it ready for revolutionary mass-action” (Bjørgum 2017, 45).
The 1917 revolution strengthened the radical movement, especially in DnA. Following the
March revolution, Tranmæl demanded that workers organize a general strike against the war, and
arm themselves against the coming counter-revolutionary attempts by the bourgeoisie. The

12
November revolution would decisively tip the power balance towards the radicals within DnA
(Fure 1983; Sundvall 2017). The news from Russia led to major organizational and ideological
transformations almost immediately. The labor paper, Klassekampen published the Zimmerwald
declaration, urging sympathizers to “establish everywhere soldier and worker councils as your body
in the struggle for peace!” (Bjørnson 1990, 509). The radicals, led by Tranmæl, would establish and
coordinate worker- and soldier councils around the country. These organizations offered Tranmæl
a base of support outside the established frameworks of DnA and AFL. Among the resolutions
adopted at the worker councils’ national conference on March 24, 1918 was the immediate
introduction of the eight-hour day, and if the government ignored their demands, “political mass-
strikes” (Bjørnson 1990, 516-517).
By 1918, the radicals had grown strong enough to challenge the reformists. Their
recommendation to the national party meeting of March 1918 stated that DnA “must … reserve
the right to use revolutionary mass-action in the fight for the working-class economic liberation”
(DnA 1918). The reformists lost at the 1918 meeting, and the radicals gained majority. Tranmæl
became party-secretary, with all reformists relinquishing their place in the party-leadership.
In November 1918, the new party leadership believed a revolution in Norway was possible,
and several members even started working on (subsequently abandoned) plans on “arrangements
for a quick takeover of power” (Fure 1983, 473). Planning group notes show that a coup would be
facilitated by the massive organization of worker councils, and the immediate reforms to be
implemented once in power were socialization of means of production and the eight-hour day.
Tranmæl was designated to be “leader of the revolution” (Ibd. 473-476).
The radical line was strengthened in 1919, when the Labor party’s national convention
declared that the “party considers mass action in its various forms to be the crucial means in the
struggle for socialism” and encouraged MPs to strike and soldier councils to form and work against
mobilization (DnA 1919, 12). Importantly, DnA decided to become member of Comintern.

13
Vice-chairman Emil Stang was the single delegate representing DnA at the First Congress
in 1919, and the party joined in late 1919. At the second conference, DnA had one of the biggest
delegations: 10 delegates arrived in Moscow, including the youth organization leader, (and later
PM) Einar Gerhardsen. At this conference, Comintern’s leader Zinonev presented “Lenin’s 21
theses”. The explicit aim of Comintern was to establish organizations, communist parties, which
could function as divisions in the European civil war and survive the coming world revolution
(Sundvall 2017). This revolutionary ideology would be too much for some reformists within DnA.
In 1921, they broke out and established the Norwegian Social Democratic Party, achieving about
10 percent of the vote. These reformists would not re-join DnA until 1927.
DnA took an active role in both legal and illegal work in Comintern. A secret committee
arranged illegal transportation of Soviet propaganda materials (Fure 1983; 468-9). Party offices
were used to hide illegals traveling to and from Russia, and prominent DnA members participated
in smuggling coordinated with Moscow. There even existed a secret agreement between DnA and
Comintern, under which DnA accepted to undertake all orders from Moscow in exchange for
political, financial, and military resources (Olstad 1998, 39).
In 1922, the high court barrister and communist Ludvig Meyer was tasked to investigate
the opportunities for revolutionary action. He concluded that “Norway is on the verge of
breakdown, which can be exploited by a tax-strike and by pushing for demands that could rally the
workers against the government.” (Olstad 1998, 49). Despite Meyer perceiving a revolution as still
likely, Comintern adherents in DnA had, by this time, started pushing for a moderate line, following
signals from the third international conference. Over time, this created a rift in the revolutionary
wing, which pushed Tranmæl and followers to embrace the remaining social democrats in DnA.
Tranmæl also wanted to use the dual membership of union and party to build a revolutionary
movement “from below”. But, Comintern, with its concept of party cells, wanted dual membership
revoked, creating a major point of contention. Subsequently, DnA voted to leave Comintern in
1923. The pro-Comintern fraction left DnA and established the Norwegian Communist Party

14
(Norges Kommunistiske Parti [NKP]) in November 1923, with 13 MPs defecting. In 1924, only
six of these MPs were re-elected, reflecting the more general power shift towards reformists and
away from revolutionaries.
3.2 Elites respond to the revolutionary threat
The described developments were not lost on Norwegian economic and political elites. The
revolution of 1917 and the following power change in DnA fundamentally changed perceptions of
the labor movement and security situation in the military, business community, and among liberal
and conservative political elites. Before 1917, the military establishment and political elites shared
the opinion that military engagement in internal affairs should be avoided (Agøy 1994, 32-34). By
1918, steps were taken to set-up “risk-free” military divisions, i.e., excluding lower-class members
(Pettersen 2010), which could be mobilized during, e.g., general strikes. At various times, military
divisions and battleships would be mobilized as pre-emptive measures against strikes getting out
of hand.
In early 1918, leading cabinet members feared an outright coup (Agøy 1994; 94). The PM
Gunnar Knudsen established a secret security commission, mandated to “secure peace and order
if civilian government was brought down” (Pettersen 2010, 43). The security commission was
summoned to re-convene on November 12, 1918, following revolutionary events in Germany. The
military secret services and police were tasked to increase their surveillance and further develop
plans for successful defense against a coup.
After DnA’s entry into Comintern and with a national rail strike looming in 1920, fears of
a revolution seem to have increased further among Norwegian political elites. The new
Conservative PM Halvorsen discussed prospects of a revolution in a speech to fellow Conservative
Party MPs:
“One is expecting the hardest of civil wars (...) Our present enemy, even with their minority
position, would still be able to win in the moment [and] we cannot know whether they
intend to secure the persons of government. Edvard [Hagerup Bull] therefore said we must

15
secure a government for the nation. He proposed that Ivar Lykke and Gunnar Knudsen
should stand by with their people if anything were to set the current government out of
play. If they and their people were to meet the same fate, the director general of the finance
department should stand ready.” (Quoted in Danielsen 1984, 18; translated by authors).
This speech strongly suggests that the revolutionary threat was perceived as credible; the
Norwegian PM was, indeed, setting up lines of succession to a competitor party and the
bureaucracy, because he believed the very existence of the regime was threatened.
Several military and police measures was organized against the worker movement. Still our
focus is on the co-opting and integrative measures – the silk glove and not the steel hand hidden
within. In Appendix B, we show that the “silk hand responses” included several (proposed or
implemented) political and economic reforms, encompassing a new electoral system, subsidies for
housing, worker councils, profit sharing, arbitration regulation, socialization of industries, generous
unemployment subsidies to unions, and old-age pensions. Space constraints limit our focus here
to work-hour regulations:
3.3 Eight-hour day
Work hours were unregulated for adults in early-20th century Norway. In 1914, the Liberal Party’s
“Great Reformator”, Johan Castberg, made his second attempt to regulate work hours. A Castberg-
led commission put a proposition, outlining two alternative policy proposals, before parliament
(the Storting). The proposals included Castberg’s (and the majority of the commission’s) favored
alternative, a nine-hour workday with compensation for overtime. The minority position suggested
a ten-hour normal workday. Both proposals were dead on arrival. In 1915, an attempt was again
made to pass an act, but Castberg’s proposal underwent extensive changes in parliament, and ended
up a major disappointment for its architect. Gone was overtime compensation, daily hours were
capped to ten, and implementation was set to 1920 with major industries excluded. Neither the
Conservatives nor Liberals could support the eight-hour day, or even a nine-hour day.

16
Regarding the economic elites, the employer association N.A.F. was, at the time, fully
against any regulation of work hours in factories.
3
N.A.F. was hesitant also in spring 1918 when
commenting on a government proposal for a temporary eight-hour day. Regulating work hours
was argued to “cause so many so many difficulties, that industrial stagnation or decline must be
expected” (Petersen 1950, 366-367). Yet, there had been movement within N.A.F., as, by 1918, it
would not necessarily work against the implementation of such an act. N.A.F.’s position continued
to change with perceptions of the revolutionary threat throughout 1918. Following revolutionary
events in Finland and Germany, it would come to accept work-hour regulations both by legislation
and in collective agreements. Especially N.A.F.’s CEO, Lars Rasmussen, argued for the necessity
of meeting the workers’ new ideological orientation by other means than force (Knutsen 1994, 29-
31). In his new-year’s speech of 1919 to the board, Rasmussen outlined the dangers facing the
organization, and the possible solution – accepting the eight-hour workday:
“Previously, our organization would respond to such demands with all the means at our
disposal. But now, I believe, that we must consider, that behind these demands stand, so
to speak, all the unrest, that in our time reigns around us on this earth, and it infects also
our situation…For if we constrain this concept too much, then the pressure might become
too great. Then the development will go on without negotiations, and the result will be that
workers say: let us now seize the day, let us take power. Then we would be stuck in a societal
upheaval, a situation that we would, by all means, seek to avert; we must be aware our times,
we have to see its signs and learn its demands. We must therefore renege on some of our
old principles (…) We must make sure that we can save what we can save.”
4
By 1918-1919, also the political elites had shifted positions. Both the Liberals and Conservatives
had voted down Castberg’s nine-hour working day proposal in 1915. In 1919, all parties would
3
Ot. prp nr 34 (1914, appendix 4, 18)
4
NAF page 4 Sentralstyre 12.01.1919. Translated by authors.

17
embrace the eight-hour day. Gunnar Knudsen (PM, Liberal) decided at the end of 1918 that it was
necessary to pass an eight-hour bill to appease the socialists. The social minister Berg (Liberals)
opened the new parliament in 1919 by stating that “we have great demands for social reforms, and
the greatest task, in my opinion, is that labor now takes precedence in our country. (…) Capital
should be a servant and helper for labor, but not its master. It is this which is the demand of our
time”.
5
When, on 14 June and 2 July 1919, the eight-hour act was put forward to the two chambers
of the Storting, it passed both in the lower (Odelsting) and higher chamber (Lagting) by acclamation.
The Conservatives in opposition were equally supportive. MP Klingenberg stated that, “we
will now approve with law a demand that workers in the whole world for a lifetime has declared to
be one of the most important to (...) achieve the social conditions under which they want to live
and have a right to live under”
6
. The extent to which both Liberals and Conservatives supported
the eight-hour day is remarkable in light of the staunch opposition just four years prior. This sudden
change of heart was noticed by the socialists. Nygaardsvold (DnA) would lament that, suddenly,
all parties across the ideological spectrum had come to embrace what they had so vehemently
fought against just four years ago:
“The road to legislative reform has been hard to travel. Each time the demand of the
workers for an eight-hour normal working day was brought forward to the Storting, the
demand was voted down, or the reform was so distorted that it would have no meaningful
impact for the workers (...) Workers therefore had to take on the issue themselves (...) I
want to add, that there is no single issue that has, to such an extent, made workers lose their
faith in the parliamentary line, that parliamentary action work (...) As long as workers did
5
1919 Stortingstidene efterm. 26 march page 646. Translated by authors.
6
1919 Stortingstidene 19 june page 141. Translated by authors.

18
not put any major force behind their demand, the Storting voted down all proposals to
reduce working-hours to 8 hours”.
7
3.4 Summary
The Bolshevik Revolution helped spur different repressive and appeasing policy changes in
Norway between 1918 and 1923. Following our expectations, Labor’s international organizational
linkages were an important factor behind Norwegian elites ascertaining a high level of revolutionary
threat. In reports and discussions, membership in international organizations such as Comintern
were used as indicators of revolutionary sentiment among workers and their organizations.
Various sources and pieces of evidence support the hypothesis that social policies were often
born out of elite fears of worker revolution. The Russian revolution and subsequent Comintern
membership fueled radicalism and radical factions in the Norwegian labor movement, and thus
the perception of labor as a radical and potential revolutionary force amongst the elites. A
combination of repressive and inclusionary tactics – especially pursuing social policy concessions
that benefitted urban workers – was developed by Norwegian elites in order to weaken radical
groups and strengthen reformists in AFL and DnA.
4. Cross-country analysis
4.1 Measures of revolutionary threat and social policies, and benchmark specification
We collected new data from various sources to construct different measures, for both our
independent and dependent variables, that are applicable across different countries and over time.
For our main measure capturing communist threat, we have drawn on several sources to code two
indicators pertaining to invitations and attendance at the first Comintern meeting. Being invited to
the Comintern presented a clear and observable signal -- anecdotally, the Comintern membership
of DnA dominated several frontpages of conservative Norwegian newspapers -- about radical
7
1919 Stortingstidene efterm. 2 July page 64. Translated by authors.

19
ideology and revolutionary motivation, given how the Russian Bolsheviks distributed invitations.
Further, we discussed how participating in the Comintern may even have had independent effects
on the actual motivation as well as capacity to organize revolution among the movements that
partook. Thus, Comintern invitations and participation should capture various features that
correlate with high perceived levels of revolutionary threat by the elites.
Regarding our specific measures, these are, first, an indicator for whether a union or party
from the country was invited to the first Comintern meeting, and, second, an indicator for whether
a party or union from the country attended this meeting with voting rights. We use the invitation
measure, which is more exogenous by not hinging on the active choice of domestic unions or
parties to partake in the meeting, as our main measure. The participation measure is used for
robustness tests. For details on measurement and sources, see Appendix A1. The main reason for
using such simple dichotomous specifications is that it is hard to judge how much extra weight one
should put on additional representatives being invited to Comintern, and how to normalize
according to, e.g., population size. The two consultative delegates from China at the first
Comintern meeting was obviously more than the one such Bulgarian delegate, but it is not obvious
that this indicates “greater representation” of China, given the large difference in population
numbers. Thus, we use the simple dummy variable set-up, but highlight that our main results are
robust to using continuous measures on number of attendees (see Appendix A6 and A8).
We are the first to acknowledge that the Comintern-based measures are not direct measures
of elite perceptions of revolutionary threat, and the measures also provide limited time series
variation for each country. (Concerning the latter, we highlight that we use different strategies to
account for this limited variation and control for other potentially relevant factors that also tended
to change around the same time period). Yet, reliable, alternative measures for revolutionary threat
are, unfortunately, hard to come by. To our knowledge, no encompassing dataset on political strikes
or communist party membership exists for the period leading up to WWI. If our Comintern
measures should be associated with a lot of noise, the most likely outcome is that regression

20
coefficients for this independent variable are attenuated, making it harder to find support for our
theoretical expectation. (In bivariate regressions, unsystematic measurement errors in regressions
always lead to attenuation bias, but exceptions might occur in multivariate analysis).
One alternative indicator of Bolshevik inspired political mobilization is the presence of
worker- and soldier councils; the Zimmerwald declaration following the Russian Revolution
explicitly encouraged the formation of such councils to spread the revolution. Hence, we collected
new data also on worker- and soldier councils, creating a dummy variable that registers if a country
experienced the formation of at least one council during 1918-19.
8
The data sources, even for
countries with extensive council formation, do not allow us to register when these councils became
defunct, but they tended to ebb out by the early 1920s (see Nordvik 1974). We use this measure in
robustness tests for our main specifications using the Comintern measure. Convergent validity tests
with the worker- and soldier council measure also suggest that our main Comintern measure is a
valid indicator of revolutionary threat. Only four countries with Comintern invitations did not have
council movements – Brazil, Turkey, Portugal and Spain -- whereof the two former had pervasive
riots, general strikes and protests organized by syndicalist unions between 1917-1920.
Our argument highlights that elites are likely to target policy benefits that mitigate
grievances among the urban working classes. For our dependent variable, we therefore need
measures that capture policies that benefit this group. To ensure robustness and that results are
not specific to a particular policy we develop and test various measures. For our first dependent
variable, we have coded a measure for “normal weekly working hours” for factory (industrial)
workers, defined as the number of hours an employee can work before overtime restrictions come
8
The often-used Banks data on general strikes is based on one source (New York Times) and does
not capture lockouts or strikes per se, only sparingly general strikes. To illustrate, in 1919 Norway
experienced the greatest strikes and lockouts waves in its history to that point but is assigned 0
strikes in Banks data. Even more concerning, the general strikes in 1920 and 23 is not captured.

21
into play. Absent regulation, we code 72 hours per week (12 hours*6 days); but results are robust
to alternative specifications. For sources used and details on this measure, e.g. on handling sectoral
differences within manufacturing, see Appendix A2.
Our alternative measures pertain to the coverage or generosity of social transfer programs
that were key to mitigating work-life risks. We have highlighted the particular threat stemming from
the urban industrial workers, and we thus focus on coverage for this particular group in our main
analysis. Thus, our second measure captures the extent to which industrial workers are protected
against central risks; old-age, sickness, maternity, accidents, unemployment, and child-rearing.
Coverage for a specific risk can be issued through various programs. Our measure takes into
account the different provisions that are possible, and that multiple programs tend to increase
coverage (within the relevant group). Yet, given the differences between programs in likelihood of
providing efficient coverage and insurance for low-income groups, our measure only considers
four types of redistributive programs (social insurance, voluntary state subsidized insurance, non-
contributory with means-testing, non-contributory without means-testing), and thus excludes non-
redistributive programs (mandatory private accounts, employer liability, lump-sum provident
schemes). The theoretical range of this measure is 0 (industrial workers covered in no program) to
24 (industrial workers covered for all 6 risks in all four relevant types of redistributive programs;
6*4=24), but the empirical range is far more limited in our main analysis below (typically 0-5).
9
Yet, social transfer programs may formally cover social groups without channeling
substantial resources to recipients. Thus, our final type of dependent variable captures the
generosity of core welfare programs, measuring duration of benefits for sickness and
unemployment, respectively. Unfortunately, these measures are only coded for the year 1925 and
we present these results in Appendix A11. Briefly, these tests paint the same picture as tests on the
9
We also test measures for whether industrial workers is covered in redistributive programs for
each risk, separately (see Appendix A7).

22
other dependent variables; countries that faced high revolutionary threats had more generous
programs than other countries.
Concerning design and model specifications, we highlight at the outset that our results are
robust to using cross-section or panel data, different estimation techniques, sets of controls, error
correction methods, etc. For our benchmark, we opt for a simple OLS model. For panel
specifications, we cluster errors by country and prefer a dual fixed effects specification, including
dummies for both countries and years. This specification eliminates several hard-to-measure
confounders that may simultaneously affect both revolutionary risk and social policy provision and
design. For example, some countries, for historical or geographical reasons, could have its industrial
production concentrated in certain sectors that both facilitate the coordination of strong unions
and provide an impetus for broad social policy coverage. In the benchmark, we also include logged
GDP per capita (p.c.) to account for level of development. GDP per capita should also correlate
with productivity, and even extent of industrialization. The GDP p.c. estimates are from Fariss et
al. (2017). These imputed data, with extensive time series, are produced from a dynamic latent trait
model on various data sources, and mitigate measurement errors of various kinds. We further
control for log population (also from Fariss et al. 2017) to account for differences in revolutionary
threats and social policies among large and small countries.
In alternative specifications, we include country-specific trends or a lagged dependent
variable. We also use the synthetic control method developed by Galiani and Quistorff (2016), first
with several treated cases and second with the usual single treated case (Appendix A10). Results
are also supported when using an instrumental variable design (using signatories on the
proclamation of the Zimmerwald group as instrument for Comintern invitations; see Appendix
A12).
4.2 Cross-national relationships
For starters, we find a strong and robust cross-sectional relationship between “revolutionary fear”,
as proxied by Comintern invitations, and working hours, as indicated by the cross-section

23
regressions in Table 1. Model 1 shows how revolutionary fear correlates strongly, and in the
expected direction with working hours, suggesting that Comintern countries had about 10.8 fewer
working hours per week than other countries. However, our measures of revolutionary threat is
also correlated with level of development and population, and in Model 2 we therefore enter log
GDP p.c. and log population as controls. The predicted difference in hours then increases to 11.4.
In extension, we use the technique developed by Frank & Xu (2018), we check how many
observations must be miscoded for this result to be invalidated. More than half of our countries
(45 of total 87) would have to be replaced with countries for which the null hypothesis (i.e., no
relationship between revolutionary threat and working hours) is true for the result to turn
insignificant. We can therefore be quite confident that this result is not driven by any individual
countries or some minor (or even major) measurement error.
Table 1. Revolutionary fear in 1919 and legislated normal working hours (DV)
(1)
(2)
(3)
Invited Comintern
-10.8***
-11.4***
(-5.03)
(-4.16)
Council Movement
-9.36***
(-3.47)
Controls
No
Yes
Yes
Countries
87
87
87
R2
0.229
0.246
0.175
Mean hours
(min-max)
66.03
45-72
66.03
45-72
66.03
(45-72)
*p<0.05, **p<0.01, ***p<0.001. t statistics in parentheses. Restricted to countries for which we have data on main
controls. Results in Model 1 robust to including whole sample. Control variables (log GDP per capita, log population)
are omitted from table.
The relationship also turns out to be persistent. In Model 3, we replicate Model 2, but include our
measure of worker- and soldier-council movement. The result is substantially very similar.
10
As we
will show when including over-time variation, this relationship turn out very robust to choices of
temporal lags, time period under consideration, measurement specifications, control variables, and
estimation techniques.
10
Results are robust to using standard errors correcting for small sample bias or jackknifed errors.

24
To illustrate the persistence of the relationship further, Figure 1 describes the development
of average working hours, across time, in countries that received Comintern invitations and those
that did not. The figure adds another line for Comintern- countries, but excluding those that
became independent after 1919. We observe an early divergence appearing between the Comintern-
invited and other countries, but that their trends are still fairly similar pre-1917; the average
difference in work hours is only three hours in 1916. This difference increases to six hours in 1917,
the year of the Bolshevik revolution, and jumps further to 12 hours in 1919, the year Comintern
formed. This difference is still 12 hours in 1923, after which it gradually declines. A moderate
difference of about 3-5 hours persists, however, to the end of the Cold War. In Appendix A14, we
highlight how this persistent, long-term relationship is mediated through Comintern participation
leading to formation of new communist parties. Where new such parties formed, lower working
hours for industrial workers persisted even after the original threat of contagion of the Russian
revolution dwindled.
Figure 1. Average working hours in manufacturing for different groups of countries.

25
One concerns is that our estimates are driven by us setting countries with non-regulated working
hours to 72 hours. We therefore carried out alternative regressions where we measure instances of
reform as dependent variable, coding the dummy 1 if a country introduces a working time law in
1919 that reduce the number of hours. We also construct another dummy measure for which
countries with any law giving a reduction in hours in 1919 are coded 1. (We checked the actual date
of each legislative change in 1919 and all changes passed parliament, or were decreed, after
Comintern invitations were issued.) We then run cross-section OLS models on this dummy (LPM),
with our Comintern invitation measure as covariate. The results, which, we highlight, are similar
for Probit models, follow theoretical expectations; a country facing revolutionary threat in 1919 is
about 40 percent more likely to introduce a working time law that reduces hours, and the results
are stable to excluding (Model 1, Table 2) or including (Model 2) controls for income and
population. Results are similar if we restrict our focus to countries that adopted their very first
working time law in 1919. Models 3-4 show a similar relationship when using the presence of a
worker or soldier movement in 1918-1919 to proxy for revolutionary threat.
Table 2. Invitation to the Comintern, Formation of Worker- and Soldier Councils, and
introduction of working hour law with reduction in work hours in 1919
(1)
(2)
(3)
(4)
Comintern Invitation
0.41***
0.41***
(7.82)
(4.35)
Council movement
0.35***
0.31*
(3.70)
(2.62)
Constant
0.017
0.037
0.359
-0.432
(0.76)
(0.06)
(1.74)
(-0.79)
Controls
No
Yes
No
Yes
Countries
146
87
146
87
R2
0.298
0.269
0.233
0.207
Mean dep.var
0.089
0.149
0.089
0.149
*p<0.05, **p<0.01, ***p<0.001. t statistics in parentheses. Linear probability (OLS) model. “Controls” are log GDP
p.c. and log population.
4.3 Main results
We now turn to our panel tests and once again employ our continuous measure of work hours and
Comintern invitation proxy of revolutionary threat. Regarding sample-composition, we exclude
Russia from all regressions. We study the threat of revolution, and not the effect of having a

26
Communist regime. Second, we run-robustness tests excluding the newly formed (post-WWI)
East-European and Baltic states; these countries enter our sample before they become independent
countries, without national legislation regulating working hours.
11
We also tested various starting-
years for our analysis, from 1789 to 1914. Since our main results remain robust to choice of starting
year (Appendix A9), we use the first year with data in most analysis, in order to avoid selection bias
and use as much information as possible to obtain good estimates on the “pre-trends”. As more
information is included, the easier it is to estimate whether the revolutionary threat in 1919 did
indeed result in a clear and persistent break from historical trends. We also use a global sample for
our benchmark – the revolutionary threat from Communist revolutions had global reach -- but we
highlight that restricting our tests to European countries yield similar results.
Table 3 presents results from our benchmark OLS panel regressions. Model 1 includes log
GDP per capita and log population as controls, but not the country- and year dummies. This model
sets the sample end year to 1925, about two years after we surmise that the revolutionary threat
stemming from the Bolshevik Revolution subsided in our case study. The estimated relationship is
somewhat higher than what the descriptive over-time trends in Figure 1 suggested. Specifically, the
Comintern invitation dummy – which can first be scored 1 in 1919, but is then scored 1 until the
time series end in 1925 for the relevant countries – is -14.4 (hours/week), and statistically significant
at all conventional levels. How sensitive is this finding to bias? Our results would be invalidated if
72.1% of the observations (6015/8348) were replaced with “null-hypothesis observations”. This
suggests that any bias must be very sizeable to invalidate the result.
Yet, Figure 1 showed differences between the countries that were invited and not invited
to Comintern also before 1919 – presumably delegates from countries that were considered ripe
for revolution were more likely to receive an invitation by Trotsky, and these countries may also
11
Results are robust to re-coding the Baltics and Poland to have the proscribed hours of the Russian
factory act of 1897, and Czechoslovakia the Austrian factory act of 1885.

27
have been inherently more likely to observe strict working hour regulations. Moreover, 1919 might
simply have been a fluke year, with across-the-board reductions in working hours for reasons
unrelated to revolutionary threats. In Model 2, we thus add country and year dummies (fixed
effects) to mitigate such confounding. The Comintern coefficient is attenuated to 10.8 hours, which
is more in line with the descriptive evidence, but remains highly significant (t=-4.4). The coefficient
and related t-values are close to identical when we enter country-specific time-trends on working
hours in Model 3. In Model 4 we also add a lagged dependent variable as regressor and difference
the dependent variable, thereby estimating a restricted Error Correction Model. The estimated
long-term coefficient indicates a substantial reduction, -15.8 hours/week (t=-4.3).
Table 3. Invitation to the Comintern 1919 and legislated normal working hours
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Dep var. measurement:
Levels
Levels
Levels
Changes
Levels
Levels
Levels
Levels
Invited Comintern
-14.41***
-10.81***
-10.71***
-2.37*
-8.38***
-7.45***
-8.41***
-10.87***
(-7.01)
(-4.43)
(-4.56)
(-2.86)
(-3.67)
(-3.69)
(-3.75)
(-3.97)
Controls
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Fixed Effects
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Country Trends
No
No
Yes
Yes
Yes
Yes
Yes
Yes
LDV
No
No
No
Yes
No
No
No
No
Observations
8348
8348
8348
8267
9592
15996
2100
9749
Countries
105
105
105
105
105
169
92
156
Start year
1789
1789
1789
1789
1789
1789
1900
1900
End year
1925
1925
1925
1925
1939
1988
1925
1988
R2
0.301
0.504
0.680
0.901
0.763
0.864
0.830
0.825
Mean hours
(min-max)
71.23
(45-72)
71.23
(45-72)
71.23
(45-72)
-0.092
(-27-3)
69.77
(40-72)
62.70
(38-72)
69.07
(45-72)
56.78
(38-72)
*p<0.05, **p<0.01, ***p<0.001. t statistics in parentheses. OLS with standard errors clustered by country. Country-
and year dummies, LDV (lagged dependent variable) and controls (log GDP per capita, log population) are omitted.
In Model 5, we expand the time series to 1939, and in Model 6 to 1988, thus also capturing the
longer-term relationship between Comintern invitations and working hours. This reduces the point
estimate (and t-value) somewhat, indicating that there is some catch-up for the “non-treated” cases
towards the end of the time series. Yet, the Comintern coefficient remains at -7.5 hours/week (t=-
3.7), even in Model 6, which extends to 1988. Finally, in Models 7-8, we set the time-series start

28
year to 1900, and re-estimate models 3 and 6 (end years 1925 and 1988, respectively). While this
reduces the estimated coefficient (in Model 7) compared to Model 3 somewhat, it increases the
coefficient and the precision of the estimate (in Model 8) when compared to Model 6. Nonetheless,
the main point is that the relationship between the Comintern measure and working hours is very
robust to time series specifications.
We conducted several other robustness tests (Appendix A5-A9), and find similar results,
for example, when using attendance (instead of invitation) at the first Comintern meeting. Results
are also robust to various sample restrictions, such as only including European countries or only
countries that were independent states over the whole period (Appendix A9).
One remaining worry – despite controls for linear country-trends and country-fixed effects– is that
countries invited to Comintern were already on a particular development path, towards fewer
working hours, just before they received the “treatment” in 1919. To mitigate this possibility we
conduct a series of placebo-tests, artificially assigning Comintern invitations to the same countries,
but in years prior to 1919. If these countries were already on a particular development path, these
“placebo-year” Comintern dummies should remain significant predictors of lower working hours.
Figure 2 draws on estimates from Model 4, Table 3 with fixed effects and country-trends, but now
re-estimated with Comintern participation artificially set to different years (1889, 1899, 1909, 1914-
1918). The results are striking; we only find a significant (t=-3.40) coefficient at conventional levels
when measuring the treatment in 1919, and point estimates are also far smaller for any other year.

29
Figure 2. Placebo tests assigning Comintern invitations artificially to years prior to 1919 and re-
estimating Model 4, Table 3.
Still, there could be factors that affect working time, correlate with revolutionary fear, and
occur/change at the end of WWI. In Table 4, we enter controls that capture such potential
confounders. These models extend Model 2, Table 3, with country- and year-fixed effects. (Adding
country-trends does not substantially change the results).
One important alternative explanation of social policy change is mass mobilization for war
(Obinger & Petersen 2017; Scheve and Stasavage 2016). States with vast mobilization of the
population in inter-state war may face strong post-war demands for social policies that redistribute
benefits from elites to the social groups that sacrificed lives and limbs in the war effort. This is
particularly salient for us, since the Bolshevik Revolution and Comintern coincided with the end
of WW1. We account for the mass mobilization explanation in our panel regression set-up by
adding a measure of percent of the population serving in the armed forces to the benchmark, in

30
Model 1, Table 4.
12
Yet, the Comintern coefficient only drops from 10.8 to 10.0 when including
this control, and remains significant at 1 percent.
Table 4. Controlling for the most likely alternative explanations on working hours
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Comintern invited
-10.0**
-9.95**
-9.74**
-9.14**
-13.2***
-10.53***
-13.36***
-9.85***
(-3.32)
(-3.35)
(-2.97)
(-3.39)
(-3.86)
(-4.28)
(-4.29)
(-4.01)
Mobilization
-0.047
(-0.27)
ILO member
-1.63
(-0.43)
Inflation
-0.0018**
(-3.38)
Social dem.
-0.95
minister in gov.
(-1.36)
Union density
0.11
(0.81)
Regime support
-2.17*
workers
(-2.84)
Social democratic minister
4.74
(0.99)
Electoral democracy index
-7.34*
(-2.82)
Controls
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Fixed effects
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Observations
3778
3778
3878
6074
560
7552
87
7046
Countries
78
78
39
63
25
98
87
101
R2
0.674
0.675
0.559
0.527
0.775
0.496
0.261
0.552
Mean hours
(min-max)
70.62
(45-72)
70.65
(45-72)
70.96
(45-72)
71.08
(45-72)
65.62
(45-72)
71.25
(45-72)
66.03
(45-72)
71.11
(45-72)
*p<0.05, **p<0.01, ***p<0.001. t statistics in parentheses. OLS with standard errors clustered by country. Time series
from 1817-1925 since we lack data on relevant covariates before 1817. “Controls” omitted from table are log GDP
p.c. and log population. Model 7 is restricted to 1919 due to lack of minister data beyond 1919.
This result is supported by a more qualitative reading of our material. In Appendix A4, we
provide summary tables of developments in various social policy areas, both for European
countries that partook in WW1 and other European countries. Comparative considerations suggest
that social policy expansions were not determined solely by mass mobilization in WW1, with several
non-warring countries that still faced revolutionary threats, such as the Netherlands and the Iberian
and Scandinavian countries, expanding social policy.
We also account for other plausible alternative explanations in Table 4. In Model 2, we control for
membership in the International Labor Organization, which (as Comintern) is established in 1919,
12
We experimented with various ways of measuring mobilization, following Scheve and
Stasavage (2016). Results are robust.

31
and had the 48-hour week as a primary goal. Rising prices could also drive labor militancy and
increase the demand for reform: the large food riots in Europe and Japan in 1917 originated in
rising food prices. We therefore enter a control for annual inflation (taken from Coppedge et al.
2018) in Model 3. Our benchmark results could also be capturing the policy-effects of workers
gaining hold of political power, independent of any revolutionary threat. Models 4-7 control for
reformist labor movements, including a dummy for the presence of a social democratic party in
Model 4 and share of workers organized in a trade union (Rasmussen and Pontusson 2018;
interpolated) in Model 5. We control for urban industrial workers being part of the regimes’ support
coalition in Model 6, using data from V-Dem v.9 (Coppedge et al. 2018) which should account for
more direct worker influence over government. We also control for whether a social democratic
minister was in government in 1918-1919 in Model 7 using only cross-sectional variation since the
minister measure is time-invariant. In Model 8, we include V-Dem’s electoral democracy index
(Polyarchy). While these two final controls mitigate confounding related to differences in regime
characteristics driving both revolutionary threat and policy initiation, including them probably also
introduces some level of post-treatment bias; revolutionary threat may be affected by, e.g., the
expansion of suffrage to the working classes (Acemoglu and Robinson 2006). Nonetheless, the size
and significance of our Comintern measure is robust to controlling for the above-mentioned
alternative explanations.
In Appendix A5, we display specifications that account for yet other plausible alternative
explanations. First, we further probe the potential role of democratic institutions, by focusing on
suffrage extension (from Coppedge et al. 2018). In addition to total suffrage, we control for female
suffrage, which may be especially relevant in our case, given possible gender differences in
preferences for various public policies (Bertocchi 2011). Second, we control for regional averages
on the dependent variable to account for emulation, learning and other mechanisms of policy
diffusion from neighboring countries (Weyland 2009). Third, since most Comintern invitations
were sent to European countries, we interact a Europe dummy with a pre-/post-1919 dummy to

32
control for the possibility that other developments in Europe drive policy shifts and generate a
spurious result. Our main result is persistent also in these specifications, although the latter test
yields insignificant results (at conventional levels) outside Europe when we include country-fixed
effects. Nonetheless, the relationship is quite robust, overall.
4.4 Welfare state coverage
Turning to our alternative outcomes, in Table 5, we investigate how revolutionary threat, as
reflected in Comintern invitations, influenced the coverage of industrial workers in redistributive
welfare programs. Given the (count) nature of these dependent variables, we estimate fixed effects
negative binominal count models in addition to OLS models, and results are robust. The OLS
results are presented in Table 5.
Table 5 Invitation to the Comintern 1919 and industrial workers coverage in redistributive
programs up to 1925 for 105 countries
(1)
(2)
(3)
(4)
(5)
Comintern invited
2.65***
2.40***
1.74***
2.10***
1.46***
(11.59)
(10.26)
(8.61)
(8.86)
(7.09)
Controls
Yes
Yes
Yes
Yes
Yes
Fixed Effects
No
Yes
Yes
Yes
Yes
Country Trends
No
No
Yes
No
No
Observations
8350
8350
8350
3898
2103
Countries
105
105
105
96
92
Start year
1789
1789
1789
1870
1900
R2
0.403
0.585
0.746
0.688
0.870
Mean welf. prog.
(min-max)
0.158
(0-5)
0.198
(0-5)
0.198
(0-5)
0.327
(0-5)
0.555
(0-5)
*p<0.05, **p<0.01, ***p<0.001. t statistics in parentheses. OLS with standard errors clustered by country. Time series
extend from 1817-1925. “Controls” are log GDP p.c. and log population.
The results in Table 5 follow our expectations. A higher revolutionary threat, as reflected
in Comintern invitations, is associated with industrial workers being covered in a greater number
of redistributive programs, and estimates are quite consistent when we omit (Model 1) or include
(Model 2) country- and year-fixed effects, and when we add country-specific trends to the fixed
effects (Model 3). Even in the latter, very restrictive model, receiving Comintern invitations in 1919
is associated with an increase in the redistributive program index of about 1.5 programs. As shown
in Models 4 and 5, results are also robust to setting the time series start year to 1870 or 1900.

33
In Appendix A7, we disaggregate the dependent variable and only look at coverage of industrial
workers in redistributive programs for specific risks (unemployment, sickness, etc.). Strikingly,
revolutionary threat is positively associated with welfare expansion independent of which labor
market or life-course risk we focus on.
5. Conclusion
We have developed a theoretical argument on how revolutionary threats that elites perceive as
credible spur these elites to react and counter this threat by providing concessions in the form of
social policies. We focused on the Bolshevik Revolution of 1917, and the subsequent formation of
Comintern in 1919. We have discussed how labor parties or unions from a country being invited
to Comintern enhanced the revolutionary threat level by radicalizing the relevant movement and
providing it with infrastructural and monetary support. Further, Comintern invitations function as
information signals to elites, leading the latter to update their beliefs on how credible the
revolutionary threat is. Elites respond with sticks and carrots to counter the threat, leading, e.g., to
expansions in social policies meant to co-opt labor.
We presented an in-depth case study of Norway after the Bolshevik revolution and various
statistical tests to assess the extent to which revolutionary fear, as captured by our new measure on
being invited to Comintern, drive social policy development. In brief, the evidence in support of
our argument is clear and robust; fear of revolution reduce working time and the coverage and
generosity of various social transfer programs, and the effects seem persistent. Notably, we find
that the revolutionary threat results persist even when accounting for alternative explanations such
as mass mobilization or social democratic or trade union presence.
This study contributes to different literatures. The literature on diffusion of revolutionary
threats, where revolutionary events in neighboring countries is used to proxy for domestic threats,
considers links to suffrage expansion and inequality (e.g., Sant’ Anna and Weller 2020). However,
our study focuses on policy-making. We are the first to investigate the role played by the Bolshevik

34
revolution on the adoption and extension of various labor market- and social policies by drawing
on comprehensive cross-country data material. Second, our novel theoretical contribution comes
from combining a theory of how elites use social policy for co-optation purposes, with an argument
on how elite perceptions of revolutionary threats are formed and influenced by international
linkages to domestic groups. Other contributions study related outcomes. Obinger and Schmitt
(2011), for example, focus on regime competition and not revolutionary threat per se. We provide
more direct and general evidence of domestic revolutionary threat fostering social policy
development. We thus add to our understanding of political factors shaping the origins of welfare
states, and thus, potentially, policy determinants of redistribution and inequality (Scheve and
Stasavage 2016; Milanovic 2016; Scheidel 2017).
Finally, we contribute to the debate on how revolutionary, as opposed to reformist, labor
movements shaped social policy development (e.g., Lipset 1983; Korpi 2006; Paster 2013). In the
absence of revolutionary threats, social policy was more likely to be stuck on a path in which
working time went unregulated. The growth of social democratic parties, as such, was often
insufficient to alter this trend. But the Bolshevik revolution, with the accompanied rise of
communist parties, was a major disruptive force. Under these conditions, also reformist social
democrats could push for concessions, using the threat of the revolutionaries as a bargaining tool
with extant elites. Future analysis of the historical role played by socialist and social democratic
parties, on different types of policy developments, should thus consider and elaborate further on
the role that this major international event played in altering these parties’ strategies and prospects
in succeeding with their policy demands.

35
References
Acemoglu, D. & J.A. Robinson (2006). Economic Origins of Dictatorship and Democracy. Cambridge:
Cambridge University Press.
Agnew, J. & K. McDermott (1998). The Comintern: A History of International Communism from Lenin
to Stalin. London: Palgrave
Agøy, N.I. (1994) Militæretaten Og "den Indre Fiende" Fra 1905 Til 1940: Hemmelige Sikkerhetsstyrker I
Norge Sett I Et Skandinavisk Perspektiv. Phd. thesis. University of Oslo: Oslo
Aidt, T.S. & P.S. Jensen (2014), Workers of the World, Unite! Franchise Extensions and the
Threat of Revolution in Europe, 1820-1938, European Economic Review, 72, 52-75.
Bertocchi, G. (2011). The enfranchisement of women and the welfare state, European Economic
Review, 55(4), 535-553.
Boix, C. (2003). Democracy and Redistribution, Cambridge: Cambridge University Press.
Bjørnson, Ø. (1990). Arbeiderbevegelsens historie i Norge bd. 2: På klassekampens grunn (1900-1920).
Oslo: Tiden.
Bjørgum, J. (2017). «Det knaker i det gamle samfunds fuger og baand» Martin Tranmæl og den
russiske revolusjon. Arbeiderhistorie. 1: 43- 63.
Carr, E.H. (1979). The Russian Revolution: From Lenin to Stalin (1917–1929). London: Macmillan.
Coppedge, M., J. Gerring, C.H. Knutsen, S.I. Lindberg, S.-E. Skaaning, J. Teorell, D. Altman, M.
Bernhard, M.S. Fish, A. Cornell, S. Dahlum, H. Gjerløw, A. Glynn, A. Hicken, J. Krusell, A.
Lührmann, K.L. Marquardt, K. McMann, V. Mechkova, J. Medzihorsky, M. Olin, P. Paxton,
D. Pemstein, J. Pernes, J.v. Römer, B. Seim, R. Sigman, J. Staton, N. Stepanova, A.
Sundström, E. Tzelgov, Y. Wang, T. Wig, S. Wilson & D. Ziblatt. 2018. Varieties of
Democracy (V-Dem) Dataset. v.8.
Danielsen, R. (1984). Høyres historie: Borgerlig oppdemmingspolitikk: 1918-1940. Oslo: Cappelen.
DNA (various). Landsmøte protokoll og referat. Arbeidernes Aktietrykkeri: Oslo.

36
Ot. prp nr 35 (1914). Om utfærdigelse av lov om arbeiderberbeskyttelse i industrielle virksomheter.
Oslo.
Esping-Andersen, G. & W. Korpi (1986) From Poor Relief to Institutional Welfare States: The
Development of Scandinavian Social Policy, International Journal of Sociology, 16(3-4), 39-74,
Fearon, J.D. (1995). Rationalist explanations for war. International Organization, 49(3), 379-414.
Fure, O-B. (1983). Mellom reformisme og bolsjevisme – Norsk arbeiderbevegelse 1918–1920.
Teori. Praksis. (PhD thesis), Universitetet i Bergen, Bergen.
Fariss, C.J., C.D. Crabtree, T. Anders, Z.M. Jones, F.J. Linder, and J.N. Markowitz. (2017). Latent
Estimation of GDP, GDP per capita, and Population from Historic and Contemporary
Sources. Unpublished Working Paper.
Frank, K.A. & R. Xu (2018). KONFOUND: Stata module to quantify robustness of causal
inferences. Statistical Software Components, Boston College Department of Economics.
Galiani, S. & B. Quistorf (2017). The synth runner package: Utilities to automate synthetic
control estimation using synth. Stata Journal, 17, 834–849
Hobsbawm, E. (1994). The Age of Extremes. Abacus: London
Iversen, T. & D. Soskice (2019). Democracy and Prosperity: Reinventing Capitalism through a Turbulent
Century. Princeton University Press, Oxford
Knutsen, P. (1994). Korporatisme og klassekamp: studier i forholdet mellom Norsk Arbeidsgiverforening,
fagbevegelsen og statsmakten, 1915-1920. PhD thesis. University of Oslo, Oslo.
Knutsen, C.H. & Rasmussen, M.B. (2018). The Autocratic Welfare State: Old-Age Pensions,
Credible Commitments, and Regime Survival. Comparative Political Studies, 51(5), 659–695.
Korpi, W. (2006). Power Resources and Employer-Centered Approaches in Explanations of
Welfare States and Varieties of Capitalism: Protagonists, Consenters, and Antagonists. World
Politics, 58(2), 167-206.

37
Kuhnle, S. (1981) The Growth of Social Insurance Programs in Scandinavia: Outside Influence
and International Forces, Ch. 4 in P. Flora & A.J. Heidenheimer (eds) The Development of
Welfare States in Europe and America, New Brunswick: Transaction Press, pp. 125-150
Lipset, S.M. (1983). Radicalism or reformism: the sources of working-class politics. American
Political Science Review, 77(1), 1-18.
Martin, C.J. and Swank, D. (2012) The Political Construction of Business Interests: Coordination, Growth,
and Equality, New York: Cambridge University Press
Milanovic, B. (2016). Global inequality: A new approach for the age of globalization. Cambridge: Harvard
University Press.
NAF (Various years). Norsk Arbeidsgiver Forening: Notes from central board 1915-1920. Unpublished
archive. Oslo
Obinger, H. & Petersen, K. (2017). Mass Warfare and the Welfare State – Causal Mechanisms
and Effects. British Journal of Political Science, 47(1), 203-227.
Obinger, H. & Schmitt, C (2011). Guns and Butter? Regime Competition and the Welfare State
during the Cold War. World Politics, 63(2), 246-270.
Olstad, F. (1998). «Til siste kamp der gjøres klar». Planer om revolusjon i Norge i 1921.
Arbeiderhistorie, 35–53.
Paster, T. (2013). Business and welfare state development: Why did employers accept social
reforms?. World Politics, 65(3), 416-451.
Petersen, E. (1950). Norsk arbeidsgiverforening: 1900-1950. Oslo: Grøndal & Søns boktrykkeri.
Pettersen, M.W. (2010). Generalstaben og de revolusjonære. Revolusjonsfrykt i generalstaben under
radikaliseringen av den norske arbeiderbevegelsen i perioden 1917–1921. MA Thesis, UiO, Oslo.
Pons, S. (2014). The Global Revolution: A History of International Communism 1917-1991. Oxford:
Oxford University Press.
Przeworski, A. (2009). Conquered or Granted? A History of Suffrage Extensions. British Journal of
Political Science, 39(2), 291-321

38
Rasmussen, M.B. & Pontusson, J. (2018). Working-Class Strength by Institutional Design?
Unionization, Partisan Politics, and Unemployment Insurance Systems, 1870 to 2010.
Comparative Political Studies, 51(6), 793-828.
Rimlinger, G. V. (1971). Welfare policy and industrialization in Europe, America, and Russia. New York:
John Wiley & Sons.
Sant’ Anna A. and L. Weller (2020). The Threat of Communism during Cold War: a constraint to
income inequality? Comparative politics, 52(3), 359-393.
Scheidel, W. (2017). The great leveler: Violence and the history of inequality from the stone age to the twenty-
first century. Princeton: Princeton University Press.
Scheve, K. & Stasavage, D. (2016). Taxing the rich: A history of fiscal fairness in the United States and
Europe. Princeton: Princeton University Press.
Stortingstidene (various years). Parliamentary debates and notes. Oslo.
Ruschmeyer, D. Stephens J.D., & Stephens E. (1992). Capitalist Development and Democracy.
Chicago: University of Chicago Press.
Sundvall, E.W. (2017). Arbeiderpartiet og Klassekrigen: striden om Moksva-tesene i 1920 i en
internasjonal kontekst. Arbeiderhistorie 1:65-83
Vis, B. (2019). Heuristics and Political Elites’ Judgment and Decision-Making. Political Studies
Review, 17(1), 41-52.
Weyland, K. (2019). Bounded rationality and policy diffusion: social sector reform in Latin America.
Princeton: Princeton University Press.
Weyland, K. (2019). Revolution and Reaction The Diffusion of Authoritarianism in Latin America.
Cambridge: Cambridge University Press.

i
Online Appendices for
Reforming to Survive: The Bolshevik Origins of Social Policies
This document contains a number of online appendices for “Reforming to Survive: The Bolshevik
Origins of Social Policies”. These appendices document and describe the data collection and
measures used, presents descriptive statistics and additional robustness tests for the large-n analysis,
and a more detailed and lengthy version of the country case study of Norway from 1915-1924.
More specifically, Appendix A1 introduces our Comintern-based measures of revolutionary threat
and how they are coded. A2 provides an excerpt from the codebook on the dataset on working
hours, which we use for our main independent variable. A3 contains tables with descriptive
statistics for all the main variables used in our cross-national analysis. A4 present additional
information on relevant features – pertaining to our revolutionary threat-focused explanation of
social policy change, but also alternative explanations related to mobilization during war – for
European countries in the aftermath of WWI.
Appendices A5-A9 provide additional tables and figures with results for different robustness tests
to the empirical analysis. A5 adds additional controls to our main model: general suffrage, female
suffrage, average regional working hours, and test for Europe dependency. A6 replicates all the
main analysis from the paper, but using a revolutionary threat measure based on Comintern
attendance rather than Comintern invitations. A7 presents various robustness tests on the
specification of the time period and sample for the panel analysis, for example setting different
start-years for when the time series of the sample should start. A8 replicates our main result table
using one additional measure of our independent variable: number of attendees at the 1919
Comintern congress. A9 provides alternative sample restrictions such as excluding various
regressions and setting different start and end years. A10 provides additional information for the
synthetic control analysis. A11 replicates the welfare state coverage results for welfare state
generosity. Here we present results for our newly collected measures of generosity, the duration of
benefits periods, for, respectively, sickness and unemployment insurance programs in 1925. Since
these data are cross-sectional only, we measure our independent and control variables in 1919. This
gives us a 6-year lag. We also tried measuring all controls contemporaneously, with no substantial
change to results. Estimates suggest that countries with higher revolutionary threat, as reflected by

ii
Comintern invitations, had on average 16-30 weeks longer duration of sickness benefits and 6-8
weeks longer duration of unemployment benefits. A12 presents and discusses results from
instrumental variable regression analyses.
Appendices A13-A15 pertain to the coding of formation of communist parties and the analyses
using these measures. A12 contains the main analysis on how Comintern participation spurred the
formation of new communist parties, and how this seems to have mediated (parts of) the long-
term effects of Comintern involvement; where new Communist parties were formed, the effect of
Comintern participation on lower working hours did not disappear, but persisted for decades. A12
includes a list of all the Communist parties that we have identified and coded, alongside their
founding year. A13 includes additional results from the final analysis of the paper on the role of
Communist parties in mediating the relationship between revolutionary threat and social policy
change. A14 lists the sources used and the coding of the dummy variable measuring formation of
worker- and soldier councils in 1918-1919.
Finally, the lengthy Appendix B provides a much more detailed version of our case study of
Norway from the article, providing detailed evidence from archival material and other sources of
events, strategies, and perceptions of the different actors. In this extended version of the case study,
we also detail how the revolutionary threat shaped policy responses by both political and economic
elites in several other areas than regulation of working hours (the focus of the shortened study in
the article), including collective bargaining, changes to the electoral rules, worker-participation in
firm management, and old-age pensions.

iii
Appendix A1: Construction the Comintern measures of revolutionary threat,
Zimmerwald group measure and International Socialist Commission membership
We constructed our measures of invitation to the first and second Comintern congresses by using
a translation of the original documents from Jane Degras’ (1955) “Communist International: 1919-
1943 Documents part 1”. Information on attendees has been compiled from
http://www.marxisthistory.org/subject/usa/eam/ci-congress19delegates.html and
https://www.marxists.org/history/international/comintern/2nd-congress/delegates.html. These
webpages, in turn, draw on information compiled from John Riddell’s (1987) “Founding the
Communist International: Proceedings and Documents of the First Congress, March 1919”, and
R.A. Archer’s (1977) “Second Congress of the Communist International: Minutes of the
Proceedings”.
List of countries from which organizations were invited to the first (1919) Comintern
meeting:
Russia, Germany, Austria, Hungary, Finland, Poland, Estonia, Latvia, Lithuanian, Czechoslovakia,
Bulgaria, Romania, Serbia(Yugoslavia), Sweden, Norway, Denmark, Netherlands, Belgian, France,
Switzerland, Italian, Spain, Portugal, United Kingdom, Ireland, United States, Australia, Japan.
List of countries from which there were attendees at the first (1919) Comintern meeting
with voting rights:
Russia, Germany, United states, Austria, Bulgaria, Finland, Hungary, Norway, Poland, Sweden,
Switzerland, Estonia, France, Latvia, Lithuania.
List of countries from which there were attendees at the second (1920) Comintern meeting
with voting rights:
Russia, Austria, Belgium, Bulgaria, Czechoslovakia, Denmark, Estonia , Finland , France , Germany
, great Britain, Hungary, Ireland, Italy, Korea, Latvia, Mexico, Netherlands, Norway, Iran, Poland,
Spain, Sweden, Switzerland, USA, Yugoslavia (Serbia).
As a note to the discussion in the text on the validity of the Comintern measure of revolutionary
threats, the historical sources clearly suggest that Comintern invitations were only extended to
radical labor organizations or radical elements within social democratic parties. Reformist, social
democratic movements and organizations were avoided by the Russians sending out the invitration.
This was done with the explicit aim to avoid the fractionalization that had taken place within the
Second International. This aim outlined in the invitational text (taken from Degras 1955):
“9. Towards the social-chauvinists, who everywhere at critical moments come out in arms against
the proletarian revolution, no other attitude but unrelenting struggle is possible. As to the
'centre'—the tactics of splitting off the revolutionary elements, and unsparing criticism and

iv
exposure of the leaders. Organizational separation from the centrists is at a certain stage of
development absolutely essential.
10. On the other hand, it is necessary to form a bloc with those elements in the revolutionary
workers' movement who, although they did not formerly belong to socialist parties, now stand by
and large for the proletarian dictatorship in the form of Soviet power. Chief among these are the
syndicalist elements in the workers' movement.
11. Finally it is necessary to draw in all those proletarian groups and organizations which,
although they have not openly attached themselves to the left revolutionary tendency,
nevertheless appear to be moving in this direction. 12. In concrete terms, we propose that
representatives of the following parties, groups, and trends shall take part in the congress (full
membership of the Third International shall be open to those parties which stand completely on
its platform)”
This invitational text was then followed by a list of invited parties and union federations, which
provide the basis of our listing of countries above.
For our instrumental variables used in Appendix A12, the Zimmerwald manifesto signatories is
taken from https://www.marxists.org/history/international/social-
democracy/zimmerwald/manifesto-1915.htm and membership in the international socialist
commission from https://en.wikipedia.org/wiki/International_Socialist_Commission
Signatories of the Zimmerwald Manifesto of September 1915:
Russia, Germany, France, Italy, Poland, Bulgaria, Netherlands, Switzerland, Sweden, Norway,
Rumania.
International Socialist Commission Members from 1915:
Italy, USA, Great Britain, Serbia, Portugal, South Africa, Greece, Russia, Poland, Lithuania, Latvia,
Finland, Switzerland, Sweden, Norway, Denmark, Spain, Netherlands, Bulgaria, Germany.

v
Appendix A2: Excerpt from Hours of Work Codebook
The Hours we Work
Regulated Working Hours around the World since 1848
A dataset on normal hours of work
Collected by Lead Author

vi
Regulating Normal Hours of Work
The “Hours we Work” dataset provides information on regulated normal hours of work as
specified by national level legislation for 203 countries from 1789 to 2014 – with the first identified
law for adult males being registered in 1848. More specifically, the regulated normal hours of work
per week registered in this dataset pertain to men in manufacturing or in general industrial work,
and refer to the number of hours that a worker can work before extra working hours are counted
as overtime. Normal hours of work is therefore not necessarily the maximum hours of work that
a manufacturing worker in the country is allowed to work. The latter is also affected by the limits
set on overtime, hours worked beyond the normal limits (and which also tend to give raise to higher
wages, and which also tend to be limited to periods or high production demand or crunch). Normal
hours of work therefore gives one indication of how many hours per week a worker is likely to
work, on average, in a year.
Countries with legal frameworks for working-time institutions can be divided into two subgroups:
those were legislation is the primary vehicle for regulating hours and those were legislation only
lays out the broad frameworks, allowing for intra- or extra-governmental institutions to fully specify
the regulatory framework. In the latter group of countries, this specification is usually done through
arbitration awards as in Australia, Italy, New Zealand, by national level collective agreements as in
Denmark and Sweden, or -- in colonies -- by governors’ being granted the power to extend
regulation on hours as they see fit.
One issue with coding normal hours of work is that there is no obvious or even conventional way
of coding the regulated normal working time in the absence of a working time law. Yet, letting
these observation stand as missing would also introduce a severe selection bias, and make us unable
to test interesting hypothesis as to the origins of working time regulations. Hence, in countries that
have unregulated working hours, we have opted set their hours at 72 hours per week (12 hours of
work, 6 days a week), which is also the weekly hours prescribed in the first law regulating hours for
both women and men (France 1848). While far from a perfect solution, this alternative is preferable
to the alternative of leaving out observations before the first law, and users of the data can easily
recode these values by using alternative scores/numbers of hours for unregulated countries.
As important as it is to highlight what the measures capture, it is also important to highlight what
they do not capture. As already mentioned, normal hours of work must not be confused with total
hours of work. Divergences are likely to be lower in countries in which overtime is very expensive

vii
for employers/highly compensated, where fewer overtime hours are accepted by law, and if
employers have to apply to authorities to use overtime.
Further, users should be aware of the following restrictions in the coding: First, these data do not
capture regulation for groups such as wageworkers in the rural sector or commercial workers
(usually regulated by their own laws). Second, the measure refers explicitly to the situation for
wageworkers, not to salaried workers. This means that legislation that refers to wageworkers is
included, while legislation that regulate hours of work for salaried workers is ignored. Third, it does
not code regulations that only regulate hours for women, adolescents, and children. Prior to the
First World War, hours of work tended to be regulated only for women and children, and some
polities continued this tradition, including the various states belonging to the United States of
America and the British empire and it’s colonial dependencies. Fifth, it does not capture regulations
that only refer to workers in harsh or hazardous occupations, as this definition tends to be too
small to encompass manufacturing or industry more generally. Sixth, countries that have no federal
or national legislation wageworkers or award-systems are classified as having no legal-system for
working-time regulation. In cases in which there exists federal regulation that regulate hours not
only for federal workers, but for all firms that buy services from the federal government, regulation
are considered as national in scope and therefore coded.
Even if a country does not have national laws regulating hours, working-time institutions still exists
through historical evolved norms or collective sector or firm bargaining agreements. This is
generally the case for the USA, Canada, and the UK in the period under study (which ends in 1989).
Unfortunately, reliably data is much harder to come by for collective bargaining than on law and
arbitration decisions. Since the tendency to rely on regulation through law, as opposed to collective
bargaining or tradition, differs systematically between countries, our data suffers from a systematic
cross-country bias. This will tend to work in the direction of showing less generous working time
regulation than what might be the picture on the ground for countries with collective bargaining
traditions. Researchers should therefor take care when interpreting results, and, for instance,
control for country-fixed effects to account for such cross-country differences whenever relevant.
An alternative source of regulation can be found in the CBR dataset (Adams, Z., et al 2017), which
measures working time regulation as outlined in both in collective bargaining and legislation. Still,
collective bargaining agreements are only rarely formulated on the national level (the Scandinavian
countries being a prominent exception). Collective bargaining agreements will therefore be less
fitting to describe national-level regulation, as hours of work will tend to diverge between sectors
and even between firms within the same sector depending on the dominant level of bargaining.

viii
The period under study includes decades in which European states colonized large parts of the
world, creating vast colonial emperies, which in turn needed labour regulation if they were to
develop internal markets allowing for commerce, exchange and exploitation. Still, hours of work
regulations were non-existent in the colonies for “native” labour (as they tended to be defined by
colonial administrations or settler governments) prior to the 20th century. Only sporadic legislation
existed, usually defining aspects such as forced labour duty and elementary rights of contracts
(similar to the master and servant acts). The extension of labour law or factory regulating hours of
native workers only started taking form in the early 1910s, taking steam after the First World War
and the Washington convention of 1919. With the 1930s and especially 1940s, the French extended
the 48-hour week into their own colonial dependencies, and in 1952, they universalized the 40
hours week to all remaining colonies. There were also marked differences between the colonial
empires in the extent to which legislation developed on a case-by-case basis or by major uniform
extension of labour law. Below we provide a short (and non-exhaustive) oversight over the major
colonial laws and give brief summaries of the extent to which a specific country tended to regulate
hours of work in its colonies by piecemeal legislation or by uniform regulations.
Below, we outline the sources and the coding procedure used to classify the 203 countries covered
by this dataset.
Sources, coding procedures, and validity
For Scandinavian, Pacific, Caribbean, African and Anglo-Saxon polities, I (first Author of paper)
mainly relied on online law databases and description of legislative changes in statistical
yearbooks.
13
For the period leading up to the 20th century, I used the description of the historical
development of factory legislation by Magnusson (1919), the extensive and very impressive 1898
report on factory working time laws by Brooke (1898), and a major League of Nations (1919) report
issued on the recent developments leading up to the declaration of the eight day work day. I also
used a report issued by the American Bureau of Labor Statistics (1913) on the ten-hour working-
day around the world. From 1919, the ILO sources (that also cover non-members) become
available; the “Legislative Series” is an annual publication that translates labor legislation into
English up to the 1980s, and is therefore used to track year to year changes from 1919 onwards,
and the various reports such as the “Hours of Work” with historical sections on previous
development are used to verify pre-1930 classifications (e.g., ILO, 1935), and the major oversight
reports of 1964, 1984, 1995, 2004, and 2018 (e.g., ILO 2018).
13
Links to webpages: Pacific, African commonwealth countries

ix
In addition, I use the following databases: NATLEX and Travail to track changes in labor laws and
hours of work laws, which are then used to identify and track down specific laws in national
databases, using the sources referred to above.
Since the dataset relies on a combination of various sources, one implication is that a hierarchy was
needed in order to ascertain which information is most trustworthy in the few cases in which
sources diverged. In short, national sources have been given precedent over ILO reports. ILO
purports to measure legislation in force in a specific year, usually the year of publication of the
relevant report. However, this raises some issues. For example, since it takes time to collect
information, several countries might have changed their regulation in the publication year, leading
to incorrect classifications.
In most cases, coding the regulated hours of work is straightforward; the legislation explicitly states
the number of normal hours of work per week in the legislation. The two most common issues we
encountered were the following. First, countries may not regulate hours by week, but instead by
day. In these cases, we calculate the normal weekly hours by multiplying the daily hours with the
number of working days (which varies over time and between countries depending on regulations
mandating weekly resting days). Second, countries sometimes do not specify normal working
hours, and instead only specify maximum working hours. In these cases, we have classified normal
working hours as maximum hours. This is defensible since workers will then have to work up to
the specified limit without overtime-compensation, meaning that workers will have to rely on some
other form of regulation besides legislation to reduce hours or receive higher compensation.

x
Appendix A3: Descriptive statistics for main models
Table I descriptive statistics for 8,348 observations from model 1 table 1.
Mean
STD
Min
Max
Weekly work
71.207
3.963
45
72
Comintern
invite
0.0221
0.147
0
1
GDP p.c. (log)
7.203
0.776
5.131
10.599
Population (log)
8.185
1.618
4.289
13.081
Table II descriptive statistics for 9592 observations from model 3 table 1.
Mean
STD
Min
Max
Weekly work
69.774
6.780
40
72
Comintern
invite
0.056
0. 231
0
1
GDP p.c. (log)
7.250
0.777
5.131
10.599
Population (log)
8.185
1.566
4.289
13.171
Table III descriptive statistics for 9592 observations from model 5 table 1.
Mean
STD
Min
Max
Weekly work
69.774
6.780
40
72
Comintern
invite
0.056
0. 231
0
1
GDP p.c. (log)
7.250
0.777
5.131
10.599
Population (log)
8.185
1.566
4.289
13.171

xi
Appendix A4:Major reforms in different European countries, grouped by
participation in WWI and history as independent country.
Table IV. Policy developments/reforms following Bolshevik revolution in countries invited to
Comintern meeting in 1919 up to 1925
Country
48 hour law
or more
generous
Unemployment
benefits
Sickness
benefits
Old-age pensions
Electoral
reform
Other major reform
Existed prior to 1914 and neutral during WW1
Denmark
1919
1918-1924↑
1915
1891
1918 -
1920 -
PR
1921 – invalidity
insurance,
employment
exchange
Netherlands
1919
1916-1917↑
1913
1919↑
1918 -
PR
1919-1920 Supreme
Labor Council
established, 1921 –
Accident insurance
(improved benefits),
1922 – accident
insurance extended
to rural workers
Norway
1919
1918-1922↑
1915-1925
↑
(1923)
1919 -
PR
1918- minimum
wage for commercial
workers, 1919 -
Social assistance,
Commission on
socialization, (1919 -
work council) 1920 –
accident insurance
(improved benefits)
Spain
1919
1919↑
(1923↑)
1919↑
-
1920 – employment
exchanges
Portugal
1922
-
(1919)
(1919)
1915 –
PR
1919 – accident
insurance extended
to all employed
persons
Sweden
1919
(1920)
1891-1921
1913
1911 –
PR
Commission on
socialization
Switzerland
1919
1924↑
1925↑
-
1919 -
PR
1918-1920 - accident
insurance, 1920 –
federal labor
department
Existed prior to 1914 and participated in WW1
Austria
1918/19
(1918) 1920
1920/21
1920
1919-PR
1919 – act of
socialization, Work
council reform,
socialization
commission and act
of community
control, 1920 – work
council act, 1919-
1921 accident
benefits increased
Bulgaria
1917/1919
1925
1918/19
1924
1913-PR
1917 – factory
inspection, 1918-24
accident insurance,
1920-1923 - land
reform
Hungary
-
-
1917/19/25
1928
1920-PR
General lack of
reform.
Germany
1918/19/22
(1918)1923
1919/23
1911/25
1918-PR
1919 – collective

xii
bargain law, 1920 -
Work council
reform, socialization
commission
Greece
1920
(1922)
(1922)
(1922)
-
1918-1925 – land
reform, 1922 –
accident legislation,
Romania
-
1991
1912
1912
-
1921-1937 - land
reform
Belgium
1922
1920-25
1910/1925
1920/24
1899-PR
1919- family
allowances, labor
inspection, 1919-
1920- accident
insurance. 1922 –
employment
contracts
France
1919
1905/10
(1898)1928
1910/1928
-
1917-24 – Family
allowances, - law on
profit-sharing
schemes.
1922 – accident
legislation
Italy
1923
1919-1923
(1919-1920)
1919-1925
1919-PR
1917- accident
insurance 1918/1920
– profit-sharing and
worker participation
in management bill
and commission
United
kingdom
-
(1911)1920
1925
1911
(1908)1920/21/24/25
1919 –
female
suffrage
1918/20– industrial
courts (arbitration
system), 1923-25
accident insurance
reforms
Ireland
-
1911/1920)
1911
1909
1922-PR
-
United States
-
-
-
-
-
1918 – immigration
act (restrictions of
anarchists) 1921 –
Emergency quota act
(restriction on
migration from
Europe)
Australia
-
-
(1912)1944
1908
-
Several important
arbitration decisions
(44 hour week in
1920, reversed in
1922)
Japan
-
-
1922
-
-
1923 – minimum age
of factory workers,
and working hours
of women and
children
Existed after 1914 and didn’t mobilize as self-organized units
Czechoslovakia
1919
1921↑
1919↑
(1919)1924↑
1919 -
PR
1918- Land reform,
1920- Socialization
of mines, 1921 –
requisitioning land
for building
purposes, Works

xiii
Committees. 1924 –
sickness and
maternity benefits
(major re-
organization)
Estonia
1931
-
1917↑
-
1920 -
PR
1921 - Wage and
Hours act for rural
workers, 1922-
accident insurance
Finland
1917
1917-1921↑
1897
-
1907 -
PR
1917-1918 –
compulsory accident
insurance, 1924-1939
land reform
Latvia
1922
-
1922↑
-
1919 -
PR
1927 – accident
insurance for all
employed persons
Lithuania
1919
-
1925↑
-
1922 -
PR
-
Poland
1919
1924↑
1920↑
-
1919 -
PR
1918 - land-reforms,
1921 – Employment
exchanges, 1924 –
compulsory accident
insurance
Oversight over major laws for labor standards and transfer welfare schemes before 1925. Health and education policies excluded.
Dates in parenthesis signify moderate or quasi-temporary measures. Dash signify no-major legislation prior to 1925. Classifications
were done by tracking legislative changes in the SpaW dataset (Rasmussen 2018), which builds on the Legislative Series produced
by the ILO, and the International Labor Review (various issues published up to 1925). Land reform from Albertus (2015) and
electoral reforms from V-Dem. Temporary-war related unemployment benefits excluded. We have not included special legislation
for smaller groups of workers, such as public servants, soldiers, miners and seamen for the simple reason that the graph would be
even more unreadable. The category “Other major” reforms include insurance reforms such as invalidity, social assistance reforms
(away from stigmatizing poor relief laws), socialization acts or commissions including land-reforms, corporative arrangements (such
as work-councils), and setting up of government organizations or departments aimed at regulating the labor market. The table is
not exhaustive and is not meant to be so. It is only a rough outline of the social policy development in non-waring and waring states
that all faced revolutionary threat.

xiv
Appendix A5: Controlling for additional alternative explanatory factors: General
suffrage, female suffrage, regional average working hours (Learning, Diffusion),
and the dependence of results on European countries (Europe dummy).
Table V Invitation to the Comintern 1919 and additional controls.
(1)
(2)
(3)
(4)
(5)
Suffrage
Female
suffrage
Regional
working
hours
Europe
interaction
Europe
interaction
without FE
Comintern
-10.1***
-9.70***
-6.59*
-5.67
-10.3**
(-4.04)
(-4.02)
(-2.03)
(-1.52)
(-3.10)
Overall suffrage
-2.24
(-1.45)
female suffrage
-0.0055
(-0.22)
Regional working hours (avg.)
0.36
(1.50)
Europe (dummy)
-0.91*
-0.36
(-2.38)
(-1.03)
Comintern*Europe
-7.25
-7.77
(-1.80)
(-1.96)
Controls
Yes
Yes
Yes
Yes
Yes
Fixed effects
Yes
Yes
Yes
Yes
No
Observations
8229
2026
5430
5673
5673
R2
0.509
0.651
0.582
0.431
0.348
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. OLS with standard errors clustered by country

xv
Appendix A6: Replication of all main tables using alternative independent
(attendance at the first Comintern meeting instead of invitations) or dependent
variables (risk-specific program coverage measures).
Table VI. Attended the Comintern 1919 and Legislated Normal Working Hours
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Levels
Levels
Levels
Changes
Levels
Levels
Levels
Levels
Attended Comintern
-16.8***
-12.7***
-12.0***
-1.60
-8.74**
-8.64***
-8.28**
-10.8***
(-5.96)
(-4.10)
(-4.20)
(-1.78)
(-3.26)
(-4.37)
(-2.99)
(-4.14)
Controls
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Country Dummies
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Year Dummies
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
LDV
No
No
No
Yes
No
No
No
No
Observations
8348
8348
8348
8267
9592
15996
2101
9749
Countries
105
105
105
105
105
169
92
156
Start year
1789
1789
1789
1789
1789
1789
1900
1900
End year
1925
1925
1925
1925
1939
1988
1925
1988
R2
0.219
0.486
0.658
0.167
0.753
0.863
0.821
0.821
*p<0.05, **p<0.01, ***p<0.001. t statistics in parentheses. OLS with standard errors clustered by country. Country- and year
dummies and control variables (log GDP per capita, log population) are excluded.
Table VII Controlling for the most likely alternative explanations for 78 countries (1817-1925),
using attendance at the first Comintern meeting (1919). Legislated Normal Working Hours is
dependent variable
(1)
(2)
(3)
(4)
(5)
(6)
(7)
Attended Comintern
-9.00*
-9.09*
-11.4**
-10.8**
-3.40
-11.9***
-11.1***
(-2.31)
(-2.62)
(-2.84)
(-3.33)
(-0.64)
(-3.60)
(-3.61)
Mobilization
-0.067
(-0.40)
ILO member
-2.47
(-0.60)
Inflation
-0.00011
(-1.54)
Soc. Dem. Party
-1.19
(-1.75)
Union density
0.032
(0.24)
Regime support Workers
-2.13*
(-2.43)
Democracy index
-8.32**
(-3.18)
Controls
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Country Dummies
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Year Dummies
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Observations
3778
3778
3527
6074
560
7552
7046
Countries
78
78
39
63
25
98
101
R2
0.642
0.644
0.508
0.520
0.764
0.472
0.537
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. OLS with standard errors clustered by country -
similar results using Ar1 correction and panel corrected standard errors. Country, year dummies and control variables
excluded.

xvi
Table VIII attended the 1919 Comintern meeting and Coverage of industrial workers in
redistributive programs up to 1925
(1)
(2)
(3)
(4)
(5)
Attended Comintern
2.49***
1.95***
1.36***
1.66***
1.22**
(6.44)
(4.66)
(3.61)
(3.57)
(2.73)
Controls
Yes
Yes
Yes
Yes
Yes
Fixed Effects
No
Yes
Yes
Yes
Yes
Country Trends
No
Yes
Yes
Yes
Yes
Observations
8350
8350
8350
3898
2103
Countries
105
105
105
96
92
Start Year:
1789
1789
1789
1870
1900
R2
0.241
0.499
0.708
0.619
0.838
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. OLS with standard errors clustered by country -
similar results using Ar1 correction and panel corrected standard errors. Country, year dummies and control variables
excluded.

xvii
Appendix A7: Replication of main models using risk-specific program coverage
measures instead of working time regulation
Table IX. Invitation to the Comintern 1919 and Coverage in redistributive welfare programs for
industrial workers for specific risks.
(1)
(2)
(3)
(4)
(5)
Unemployment
Old age
Sickness
Maternity
Accident
Comintern Invited
0.20**
0.41***
0.34***
0.39***
0.49***
(2.66)
(3.76)
(3.77)
(4.38)
(5.38)
Controls
Yes
Yes
Yes
Yes
Yes
Fixed Effects
Yes
Yes
Yes
Yes
Yes
Observations
8350
8350
8350
8350
8350
Countries
97
97
97
97
97
R2
0.357
0.372
0.449
0.444
0.467
Mean welf. prog.
(min-max)
0.005
(0-1)
0.025
(0-2)
0.018
(0-1)
0.020
(0-1)
0.027
(0-1)
*p<0.05, **p<0.01, ***p <0.001. t statistics in parentheses. OLS with standard errors clustered by country. “Controls” are log GDP p.c.
and log population.

xviii
Appendix A8: Replicating main models using alternative independent variable
measurement: number of Comintern attendees.
Table X Number of Attendees at the 1919 Comintern meeting and Legislated Normal Working Hours
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Levels
Levels
Levels
Changes
Levels
Levels
Levels
bolsh_8A
Numb. Att.
Comintern
-14.1***
-8.62**
-7.53*
-0.96
-3.56
-2.65
-5.85*
-6.27*
(-5.70)
(-2.93)
(-2.58)
(-1.28)
(-1.24)
(-1.01)
(-2.30)
(-2.07)
Controls
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Fixed Effects
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Country Trends
No
No
Yes
Yes
Yes
Yes
Yes
Yes
LDV
No
No
No
Yes
No
No
No
No
Observations
8348
8348
8348
8267
9592
15996
2100
9749
Countries
105
105
105
105
105
169
92
156
Start year
1789
1789
1789
1789
1789
1789
1900
1900
End year
1925
1925
1925
1925
1939
1988
1925
1988
R2
0.232
0.464
0.641
0.164
0.744
0.861
0.819
0.820
Mean hours
(min-max)
71.23
(45-72)
71.23
(45-72)
71.23
(45-72)
-0.092
(-27-3)
69.77
(40-72)
62.70
(38-72)
69.07
(45-72)
56.78
(38-72)
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. OLS with standard errors clustered by country.

xix
Appendix A9: Testing alternative sample restrictions for main Comintern analysis
Table XI Checking Sample restrictions, all analysis ends in 1925: Regression results with working
hours as dependent variable.
(1)
(2)
(3)
(4)
(5)
(6)
Sample
restriction:
1900
Independent
and 1900
Excluding
Asia/Africa
1900
Independent
and 1900
Excluding
Asia/Africa
Comintern
Invited
-
10.3***
-8.17**
-8.54**
(-4.34)
(-2.88)
(-2.84)
Attended
Comintern
-
11.6***
-7.71*
-8.23*
(-3.76)
(-2.10)
(-2.26)
Controls
Yes
Yes
Yes
Yes
Yes
Yes
Country
Dummies
Yes
Yes
Yes
Yes
Yes
Yes
Year Dummies
Yes
Yes
Yes
Yes
Yes
Yes
Observations
2101
1388
4218
2101
1388
4218
Countries
92
66
58
92
66
58
R2
0.649
0.677
0.604
0.623
0.657
0.588
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. OLS with standard errors clustered by country - similar results
using Ar1 correction and panel corrected standard errors. Country, year dummies and control variables excluded.
Table XII Checking Sample restrictions by further variation in start year: Regression results with
working hours as dependent variable.
(1)
(2)
(3)
1905
1910
1914
Comintern Invited
-10.00***
-9.46***
-8.77***
(-4.28)
(-4.14)
(-3.96)
Controls
Yes
Yes
Yes
Country Dummies
Yes
Yes
Yes
Year Dummies
Yes
Yes
Yes
Observations
1721
1336
1023
Countries
91
91
91
R2
0.672
0.712
0.774
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. OLS with standard errors clustered by country -
similar results using Ar1 correction and panel corrected standard errors. Country, year dummies and control variables
excluded.

xx
Table XIII. Robustness test using alternative start year: Invitation to the Comintern 1919 and
Coverage in redistributive welfare programs for industrial workers for specific risks between
1900 and 1925
(1)
(2)
(3)
(4)
(5)
Unemployment
Old age
Sickness
Maternity
Accident
Comintern
Invited
0.19*
0.32***
0.22**
0.26***
0.33***
(2.47)
(3.67)
(3.23)
(3.78)
(3.74)
Controls
Yes
Yes
Yes
Yes
Yes
Fixed Effects
Yes
Yes
Yes
Yes
Yes
Observations
2103
2103
2103
2103
2103
Countries
97
97
97
97
97
R2
0.549
0.740
0.759
0.737
0.785
Mean
(min-max)
0.022
(0-1)
0.097
(0-2)
0.068
(0-1)
0.074
(0-1)
0.100
(0-1)
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. OLS with standard errors clustered by country -
similar results using Ar1 correction and panel corrected standard errors. Country, year dummies and control
variables excluded.
Table XIV. Robustness test using alternative start year: Invitation to the Comintern 1919 and
Coverage in redistributive welfare programs for industrial workers for specific risks between
1914 and 1925
(1)
(2)
(3)
(4)
(5)
Unemployment
Old age
Sickness
Maternity
Accident
Comintern Invited
0.16*
0.14
0.097
0.10
0.15*
(2.09)
(1.70)
(1.75)
(1.84)
(2.39)
Controls
Yes
Yes
Yes
Yes
Yes
Fixed Effects
Yes
Yes
Yes
Yes
Yes
Country Trends
Yes
Yes
Yes
Yes
Yes
Observations
1025
1025
1025
1025
1025
Countries
91
91
91
91
91
R2
0.738
0.907
0.907
0.902
0.910
Mean
(min-max
0.042
(0-1)
0.143
(0-2)
0.113
(0-1)
0.120
(0-1)
0.151
(0-1)
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. OLS with standard errors clustered by country -
similar results using Ar1 correction and panel corrected standard errors. Country, year dummies and control
variables excluded.

xxi
Appendix A10: Synthetic control analysis
Here we provide a short description for how we employ the synthetic control method.
We have in the regression analysis presented in the paper and previous appendices considered all
non-treated countries – i.e., those that did not have unions or parties invited to Comintern – as
relevant “contrast cases” when estimating the relationship. Still, the inclusion of less relevant cases
in the control group might affect results, despite our control strategy. Thus, we construct a more
plausible counterfactual to our treated cases by creating a set of synthetic control countries, mixing
features from several actual countries, where we try to match the pre-trend in our synthetic cases
to the treated countries. These synthetic control cases should be artificial composite “countries”
that – except for Comintern invitations – resemble the countries receiving Comintern invitations.
The treatment effect is then the difference between the actual working hours observed in our
“treated countries” and in our synthetic control cases. To aggregate the results of each individual
compilation to one final estimate each affected country treated is compared to its individual
synthetic case, which is then averaged up to a final estimate. To make this comparison substantially
meaningful we normalize by setting the working hours of each treated country to 1 in the treatment
year (1919).
To run this analysis, we first balance our unbalanced dataset to countries with no missing values
between 1900-1925 on GDP per capita and population, reducing our sample to 1898 observations
(73 countries). To construct our synthetic control, we include our two basic controls (GDP,
population) averaged over the time-period, and two pre-treatment period lags on the dependent
variable (1910, 1915).
14
The aggregated results are presented in Figure I. The averaged pre-
treatment trends for the treatment and control groups are quite similar, and Figure I shows a clear
and large, immediate effect of the treatment.
Norway was the empirical context for our qualitative case study, with this analysis suggesting a clear
effect of revolutionary threat, and Comintern involvement of the Norwegian Labor Party more
specifically, on working hours. We thus focus on synthetic control results for this particular
country. To construct a synthetic case for Norway we follow the same procedures as above (average
of controls for pre-treatment period and two time periods of the dependent variable included). The
results are shown in Figures III and IV. We see that our synthetic control case and the “actual case”
of Norway match up very well before treatment, with divergence following only after 1919. Figure
14
We experimented with additional time-periods and this generally increase the fit. At 50 % of the pre-treatment-
period, the results become insignificant.

xxii
IV shows the possible placebo-treatments, which suggests that for the first four years the observed
effect from the treatment in Norway is highly dissimilar from the placeboes.
Figure I: Effect plot of results from synthetic control analysis with multiple treated cases. Trends
in treated and non-treated group scaled to 1 in 1918 to faciliate comparision.
Figure II: P-values for treatment effect, after treatment, for the synthetic control analysis.

xxiii
Figure III: Effect plot of results from synthetic control analysis with norway 1919 as treated case.
Figure IV: Donor plot of results from synthetic control analysis with norway 1919 as treated case.
Single black line shows treated norwegian case amd white lines shows the various placebo-
treatments.

xxiv
Appendix A11: Welfare State Generosity
Table XIII Generosity of Sickness and Unemployment Benefits measured in duration of benefit
period (Weeks) in 1925
(1)
(2)
(3)
(4)
(5)
(6)
Program:
Sickness
Unemployment
Comintern Invited
24.4***
28.4***
15.9*
5.73*
7.83**
5.70*
(5.81)
(4.16)
(2.60)
(2.46)
(3.16)
(2.61)
GDP (log)
0.51
-0.73
6.13
2.25
1.28
3.86
(0.21)
(-0.16)
(1.09)
(1.24)
(0.65)
(1.20)
Population (log)
-0.89
-1.38
-0.92
0.018
0.42
0.43
(-1.27)
(-1.01)
(-0.59)
(0.04)
(0.63)
(0.35)
Mobilization
-1.18
-0.63
(-0.95)
(-1.02)
ILO member
-0.040
5.20*
(-0.01)
(2.66)
Inflation
0.0054***
0.0015**
(4.97)
(2.98)
Social De. Party
-3.84
-0.39
(-0.89)
(-0.17)
Observations/countries
84
42
29
87
45
32
R2
0.590
0.579
0.513
0.312
0.448
0.462
Weeks
7.36
(0-52)
13.19
(0-52)
11.87
(0-52)
2.50
(0-26)
3.68
(0-26)
4.35
(0-26)
*p<0.05, **p<0.01, ***p<0.001. t statistics in parentheses. OLS with standard errors clustered by country. Independent variables measured in 1919.

xxv
Appendix A12: Instrumental variable regression results
Our main specifications control for country-fixed effects and some specifications even include country-
specific trends. Yet, Comintern invitations could have been highly responsive to the perceptions of leaders
in the Russian Communist Party about which countries were ripe for revolution in the aftermath of the
dramatic and game-changing events of 1917. More specifically, Lenin, Trotsky, or other could form
judgements (based on factors that are not captured by our control variables, fixed effects, or time trends)
about which countries were experiencing short-term developments that made them particularly likely to
follow the Russian example. If so, it might be rational for Russian leaders to send Comintern invitations to
unions or labor parties in exactly these countries, anticipating that this could potentially tilt the balance of
scales between elites and revolutionaries in this country. Such perceptions, formed right before or in 1919,
may correlate also with the passage of social policies in the aftermath of 1919. In order to account for such
endogenous selection into Comintern stemming from short-term developments, we use Zimmerwald
membership – i.e., signing the proclamation of the Zimmerwald group, a break out of radicals from the
Second International in 1915 (notably, before the Bolshevik Revolution) – as a predictor of Comintern
invitations four years later. The Zimmerwald declaration, formulated by Lenin, was aimed to push worker
movements in the direction of his revolutionary line. While Lenin detested the organizational arm of the
Zimmerwald movement, the International Socialist Commission, it was a precursor to the Comintern
Agnew & McDermott 1998).
We thus use Zimmerwald membership as an instrument and Comintern invitations as the endogenous
regressor in Instrumental Variable (IV) analysis. Our identifying assumption is that historical international
worker radicalism – captured by Zimmerwald membership – can only influence current social policy through
predicting Comintern representation, once we control for current domestic worker radicalism (we try out
different such measures as additional controls to ensure the exclusion restriction is satisfied). If true, these
IV regressions will yield consistent estimates of the effect of Comintern invitations on work hours.
Results from (2SLS) IV regressions are reported in the table below. The first-stage results clearly show that
Zimmerwald membership is a very strong predictor of Comintern invitations; the specification passes all
conventional tests for instrument strength. Further, the coefficient of Comintern invitation on working
hours is larger than in our benchmark OLS specification, and various 2SLS specifications show that the
relationship remains highly significant and robust. In other words, even when we account for the possibility
that (short-term) developments in revolutionary threats is causing invitations to Comintern, Comintern
invitations are strongly related to subsequent changes in working hour regulations.
15
15
Equivalent such instrumental variable tests on other outcome variables, such as social policy coverage and generosity, show
similar results.

xxvi
Table XIV Instrumental IV regression and Invitation to the Comintern 1919 and Legislated
Normal Working Hours up to 1925
(1)
(2)
(3)
(4)
(5)
First Stage
Zimmerwald
0.513***
0.463***
0.417***
0.221
(13.75)
(7.21)
(5.90)
(1.93)
ISC
0.410***
0.307**
(5.01)
(2.74)
IV
Second Stage
Comintern Invited
-20.8***
-17.9***
-18.1**
-14.7*
-15.9**
(-4.74)
(-3.35)
(-2.95)
(-2.54)
(-3.14)
Mass. Mob.
-0.16
-0.16
-0.14
-0.15
(-1.13)
(-1.06)
(-0.92)
(-0.97)
ILO Member
-0.15
0.35
0.24
0.28
(-0.04)
(0.09)
(0.07)
(0.07)
Communism
1.09
0.13
0.46
(0.37)
(0.06)
(0.21)
Reduced Form
Zimmerwald
-10.690***
-8.318***
-7.544***
-5.271
(-4.62)
(-3.44)
(-3.03)
(-1.74)
ISC
-6.035**
-3.571
(-2.47)
(-1.28)
Controls
Yes
Yes
Yes
Yes
Yes
Fixed effects
Yes
Yes
Yes
Yes
Yes
Observations
8485
3887
3164
3164
3164
Countries
106
79
55
55
55
Kleibergen-Paap
189.005
52.008
34.756
25.077
26.893
Endogenity Test
4.222
(P= 0.039)
2.351
(P=0.125)
2.378
(P= 0.123)
1.205
(P=0.272)
2.4458
(P=0.117)
R2
0.486
0.670
0.570
0.596
0.588
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. OLS with standard errors clustered by country - similar results using Ar1 correction
and panel corrected standard errors. Country, year dummies and basic control variables excluded. Basic controls are GDP per capita (log) and
population (log) included in all models. Mass mobilization, ILO membership and communism variables included but not presented in first stage
and reduced form.

xxvii
Appendix A13: Testing a mechanism: Comintern, the formation of Communist parties
and persistent effects on social policy
Before concluding, we evaluate some additional implications of our argument. First, we ask: did Comintern
lead to the formation of communist parties around the world? Second, did formation of communist parties
matter for the persistence of the relationship between Comintern invitations and social policies documented
above? In order to respond to these questions, we collected data on the year and nature of communist party
formations, distinguishing between the formation of independent, new parties and the breakaway of central
elements from an existing social-democratic party.
16
Model 1, Table 6 reports a logit model testing whether having attended the 1919 Comintern is associated
with having a communist party, using 1940 as the time series end-point. Model 2 tests whether having
attended Comintern is associated with establishing a communist party (going from 0 to 1). In order to
include country-fixed effects, we re-estimate Model 1 using OLS as Model 3. Model 4 also adds country-
specific trends, whereas Model 5 extends the time series to 1988. Having attended the Comintern meeting
in 1919 increases the predicted probability of having a communist party of about 20 percentage points
according to the OLS results.
Table XV. Did the Comintern lead to the foundation of communist parties?
(1)
(2)
(3)
(4)
(5)
Logit
Logit
OLS
OLS
OLS
Attended Comintern 1919
4.18***
0.75*
0.20**
0.20*
0.24**
(7.54)
(2.06)
(3.02)
(2.16)
(3.21)
Controls
Yes
Yes
Yes
Yes
Yes
Fixed Effects
No
No
Yes
Yes
Yes
Country Trends
No
No
No
Yes
Yes
Observations
6752
6729
6752
6752
9649
Countries
60
60
60
60
63
End year
1940
1940
1940
1940
1988
(Pseudo) R2
0.2413
0.0259
0.758
0.807
0.883
*p<0.05, **p<0.01, ***p<0.001. t statistics in parentheses. Standard Errors clustered by country.
We also test whether Comintern attendance enhanced the frequency, more specifically, of communist
parties being formed by a split from social democratic parties. In Appendix 10, we re-estimate Table 5, but
with this dependent variable. There is a clear relationship between Comintern attendance and Communist
party formations stemming from a split of existing social democratic parties.
We have established a link between the Comintern and the institutionalization of radical worker movements
into Communist parties. Did such developments help shape social policy expansion. Figure 3 describes the
16
In Appendix A11 we list name and year of founding for all coded parties (including syndicalist parties), and whether these parties
arose by breaking from a labor party.

xxviii
development of average working hours over time for two groups of countries, Comintern countries and
others. We added a long-dash line for the average for those Comintern countries that observed organized
communist parties. Interestingly, the two sub-groups of Comintern countries track each other closely up
to the late 1930s. However, starting in the 1940s, counties with communist parties had, on average, about
1-2 hours lower working time than those without. This suggests that the long-term persistence in effect of
Comintern on social policies might partly be mediated by the formation of Communist parties, organizations
that could maintain the presence of radical forces in the country.
Figure V: Average working hours, across time, for different groups of countries.
Still, these differences could be driven by various other factors between the two groups of countries. We
therefore re-run Model 2, Table 3, but now both including and excluding the communist party variable
before we compare results. We test three sample specifications in order to assess shorter- versus longer-
term relationships. The time series ends in 1940 in Model 1, Table 7, 1960 in Model 2, and 1988 in Model
3.
The predictive power of the Comintern variable declines in all models that include the communist party
dummy. Up to 1940, the Comintern shock, in itself, is still important, as indicated by the coefficient being
significant at 5% even when controlling for the communist party dummy. In the 1960- and 1988 samples,
however, the Comintern variable loses statistical significance once controlling for communist party
formation. One interpretation is that the long-term effect of the revolutionary threat-shock from the
Bolshevik revolution and Comintern on work hours is mainly mediated by the formation of communist

xxix
parties. The communist party variable has a strong and precisely estimated coefficient on working hours in
different specifications.
Table XVI. Long-term effect of Comintern-shock when controlling or not controlling for
communist parties on legislated normal working hours
(1)
(2)
(3)
End Year:
1940
1960
1988
No communist party control
Comintern Invited
-7.07**
-4.84*
-4.56*
(-2.83)
(-2.16)
(-2.05)
R2
0.754
0.828
0.864
With communist party control
Comintern Invited
-5.84*
(-2.28)
-3.56
(-1.62)
-2.55
(-1.19)
Communist party
-3.48*
(-2.26)
-4.56**
(-2.95)
-6.79***
(-3.97)
R2
0.759
0.833
0.872
Controls
Yes
Yes
Yes
Fixed Effects
Yes
Yes
Yes
Observations
6693
7941
9649
Countries
60
60
63
*p<0.05, **p<0.01, ***p<0.001. t statistics in parentheses. OLS with standard errors clustered by country. “Controls”
are log GDP p.c. and log population.

xxx
Appendix A14: Communist parties
Table XVII Complete list of coded communist parties
Country
Communist party name
Year of founding
Sweden
Svergies Kommunistiska Parti
1924
Norway
Norges Kommunistiske Parti
1923
Denmark
Danmarks Venstresocialistiske Parti
1919
Bulgaria
Balgarska rabotnicheska sotsialdemokraticheska
partia
1903
Belgium
Parti Communiste de Belgique
1921
Finland
Suomen Kommunistinen Puolue
1919
Estonia
Eestimaa Kommunistlik Partei
1920
Latvia
Latvijas Komunistiskā partija
1904
Lithuania
Lietuvos komunistų partija
1918
Greece
Kommounistikó Kómma Elládas
1918
Switzerland
Kommunistische Partei der Schweiz
1921
Russia (Soviet Union)
Bolsheviks
1903
Albania
Partia Komuniste e Shqipërisë
1928
Poland
From 1906: Polska Partia Socjalistyczna – Lewica
From 1918: Komunistyczna Partia Polski
1906
Portugal
Partido Communista Portugues
1921
France
Parti Communiste Francais
1920
Germany
Kommunistische Partei Deutschlands
1918
Ireland
Socialist Party of Ireland
1904
Italy
Partido Communista Italiano
1921
Netherlands
From 1909: Social Democratic Party
From 1918: Communistische Partij Nederland
1909
Spain
Partido Communista de Espana
1921
Turkey
Türkiye Komünist Partisi
1920
United Kingdom
Communist Party of Great Britain
1920
Austria
Kommunistische Partei Österreichs
1918
Romania
Partidul Comunist Român
1921
Serbia (Great Serbia -
Yugoslavia)
Serbian name: Zveza komunistov Jugoslavije
Official name: Savez komunista Jugoslavije (from
1952)
English name: League of Communists of
Yugoslavia
1919
Hungary
Magyar Kommunista Párt
1918
Mexico
Partido Socialista Obrero
1917
Colombia
Partido Comunista Colombiano
1930
Brazil
Partido Comunista Brasileiro
1922
United States
Communist Party of the United States of America
1919
El Salvador
Partido Comunista
1930
Bolivia
Partido Obrero Revolucionaro
1938
Haiti
French: Parti Communiste Haïtien
Creole: Pati Kominis Ayisyen
1934
Honduras
Partido Comunista de Honduras
1927
Peru
Partido Socialista del Perú
1928
Argentine
Partido Comunista de la Argentina
1918
Venezuela
Partido Comunista de Venezuela
1931
Nicaragua
Partido Obrero Socialista
1967
Canada
Communist Party of Canada
1921
Chile
Partido Obrero Socialista
1922

xxxi
Paraguay
Partido Comunista Paraguayo
1928
Cuba
Partido Comunista de Cuba
1925
Costa Rica
Partido Comunista Costarricense
Primarily called: Bloque de Obreros y Campesinos
1931
Ecuador
Partido Socialista Ecuatoriano
Changed name to Partido Comunista del Ecuador
in 1931.
1925
Guatemala
Partido Comunista de Guatemala
1922
Panama
Partido del Pueblo de Panamá
1930
Uruguay
Partido Comunista del Uruguay
1920
Dominican Republic
Partido Comunista Dominicano
1944
Japan
Nihon Kyōsan-tō
1922
Yemen
National Liberation Front
1963
Pakistan (India)
Communist Party of Pakistan
1948
Afghanistan
People's Democratic Party of Afghanistan
1965
India
Communist Party of India
1928
Thailand (Siam)
Communist Party of Thailand
1942
Nepal
Communist Party of Nepal
1949
Australia
Communist Party of Australia
1920
Iran
Hezb-e-Komunist Iran
1920
Iraq
al-Hizb al-Shuyu'i al'Iraqi
1934
Syria
Hizb al-Sha'b al-Lubnani
al-Hizb al-Shuyu'i al-Suri
1924
1944
Tunisia
al-Ḥizb ash-Shuyū‘ī at-Tūnisī
1934
Saudi Arabia
Hizb al-Shuyu'i fial-Sa'udiyah
1975
South Africa
Communist Party of South Africa
1921
Mali
Parti Malien du Travail
1965
Ethiopia
Mela Ethiopia Sosialist Niqinaqē
1968
Liberia
--
--

xxxii
Appendix A15: Additional results for tests pertaining to communist parties
Table XVIII. Did the Comintern lead to a split from social democratic parties to
communist parties?
(1)
(2)
(3)
(4)
(5)
Logit
Logit
OLS
OLS
OLS
Attended Comintern
2.47***
1.38*
0.43**
0.45**
0.51***
(3.33)
(2.39)
(3.14)
(3.07)
(3.76)
Controls
Yes
Yes
Yes
Yes
Yes
Fixed Effects
No
No
Yes
Yes
Yes
Country Trends
No
No
No
Yes
Yes
Observations
6954
6930
6954
6954
9943
Countries
60
60
60
60
63
End year
1940
1940
1940
1940
1988
R2
0.3232
0.1146
0.479
0.679
0.866
t statistics in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001. Standard Errors clustered by country. Pseudo R2 in models 1 -2. Country, year
dummies and control variables excluded. Basic controls are GDP per capita (log) and population (log). Results are substiantialy similar if we constrain
our sample to countries with social democratic parties.

xxxiii
Appendix A16: Worker- and Soldier Councils
Here we present the sources and short descriptions of the countries that had a worker or soldier
council movement.
General sources:
Cronin, J. (1980). Labor Insurgency and Class Formation: Comparative Perspectives on the Crisis
of 1917-1920 in Europe. Social Science History, 4(1), 125-152. doi:10.2307/1170882
Map of Workers' Councils (1917-1927): link https://libcom.org/history/interactive-map-workers-
councils-1917-1927
Worley, M. (2008) The Soviet Union and Bolshevism abroad. In: Atkin, N. (ed.) Themes in Modern
Euorpean History, 1890-1945.
Rathbone, K. (2007). Coming Home: Political Radicalization in Western Europe's Front
Generation (Doctoral dissertation, The Ohio State University). Linke
https://kb.osu.edu/handle/1811/29956
Country-specific sources:
Austria
Worker councils and massive strike waves 1918 followed by soldier councils and radicalization of
worker councils 1919.
Sources:
Leidinger/Moritz, Gefangenschaft, Revolution, Heimkehr 2003, pp. 556, 570 and 582;
Maderthaner, Wolfgang: Utopian Perspectives and Political Restraint: The Austrian Revolution in
the Context of Central European Conflicts, in: Bischof, et al: From Empire to Republic 2010, p.
54-56.
F. L Carsten, Revolution in Central Europe, 1918-1919 (Berkeley: California University Press,
1972), chap. 4.
Belgium
In Brussels, a short-lived German central workers’ and soldiers’ council was established on 10
November. Attempts to fraternize with the Belgian population were opposed by the leadership of
the socialist party, who also rejected invitations to cooperate with the council.
https://encyclopedia.1914-1918-online.net/article/labour_belgium

xxxiv
Bulgaria
The first soviets (councils) of soldiers and workers began forming in Bulgarian cities on 23
September 1918. In the capital, Sofia, open rebellion was only checked by a German division sent
from Russia. Once the armistice was signed, Bulgarian forces were demobilized. All foreign troops
had to surrender or leave the country.
Sources:
https://encyclopedia.1914-1918-online.net/article/revolutions
Brazil
Only a few councils were ever formed, expect during the large scale strike of 1919
Sources:
C.Mir (2013). The Russian revolution echoes in Brazil, 1918-21. International Review no.151. link
https://en.internationalism.org/internationalreview/201305/7739/russian-revolution-echoes-brazil-1918-
21
Chile
Worker councils formed during 1918. Little information available.
Sources:
Map of Workers' Councils (1917-1927): link https://libcom.org/history/interactive-map-workers-
councils-1917-1927
Czechoslovakia
Councils formed in old-Austrian and Hungarian parts of the country.
Sources:
Same as Hungary and Austria
Denmark
Worker and soldier councils formed around the country in 1917-1919

xxxv
Sources:
Nils Ivar Agøy. (1996). When Officers Need Internal Enemies: Aspects of Civil-Military
Relations in Scandinavia between the World Wars. Journal of Peace Research, 33(4), 469-481.
Retrieved January 7, 2020, from www.jstor.org/stable/424570
Estonia
Soldier and worker councils extensively formed 1917-1919
Sources:
Andreyev, A., (1971). The Soviets of Workers' and Soldiers' Deputies on the Eve of the October
Revolution, March-October 1917. Progress Publishers.
France
Soldier and worker councils formed between 1918-1919. Soldier mutinies in 1917 and 1919 lead
by councils. Most mobilization captured by syndicalist organizations.
Sources:
Megan Trudell - Prelude to Revolution: Class Consciousness and the First World War (1997)
Bentley B. Gilbert and Paul P. Bernard - The French Army Mutinies of 1917 (1959)
Finland
Soldier and worker councils extensively formed 1917-1918
Sources:
Anthony F. Upton - The Finnish Revolution 1917-1918 (1980)
D. G. Kirby (1978) Revolutionary ferment in Finland and the origins
of the civil war 1917–1918, Scandinavian Economic History Review, 26:1, 15-35, DOI:
10.1080/03585522.1978.10407894
Germany
Soldier and worker councils extensively formed 1917-1919.
Sources:

xxxvi
Cronin, J. (1980). Labor Insurgency and Class Formation: Comparative Perspectives on the Crisis
of 1917-1920 in Europe. Social Science History, 4(1), 125-152. doi:10.2307/1170882
Kuhn, G. (2012) All Power to the Councils! A documentary history of the German Revolution
of 1918 - 1919. PM Press.
Hungary
Soldier and worker councils extensively formed 1917-1919
Source:
Cronin, J. (1980). Labor Insurgency and Class Formation: Comparative Perspectives on the Crisis
of 1917-1920 in Europe. Social Science History, 4(1), 125-152. doi:10.2307/1170882
Deak, I. (1968). Budapest and the Hungarian Revolutions of 1918-1919. The Slavonic and East
European Review, 46(106), 129-140.
Latvia
Soldier and worker councils extensively formed 1917-1919
Source:
Andreyev, A., (1971). The Soviets of Workers' and Soldiers' Deputies on the Eve of the October
Revolution, March-October 1917. Progress Publishers.
Lithuanian
Soldier and worker councils extensively formed 1917-1919
Source:
Andreyev, A., (1971). The Soviets of Workers' and Soldiers' Deputies on the Eve of the October
Revolution, March-October 1917. Progress Publishers.
Ireland
Soldier and worker councils formed 1918-1922
Sources:
O'Connor, E. (1985). Active sabotage in industrial conflict, 1917–23. Irish Economic and Social
History, 12(1), 50-62.
Kemmy, J. (1976). The Limerick Soviet. Saothar, 2, 45-52.

xxxvii
Italy
Extensive factory council movement in the north 1918-1920s
Sources:
https://en.wikipedia.org/wiki/Biennio_Rosso
Bellamy, R., Bellamy, R. P., & Schecter, D. (1993). Gramsci and the Italian state. Manchester University
Press.
Netherlands
Soldier and worker councils formed 1918 including revolution attempt by labor leaders.
Sources:
https://en.wikipedia.org/wiki/Communist_Party_of_the_Netherlands
https://en.wikipedia.org/wiki/Red_Week_(Netherlands)
Norway
Soldier and worker councils extensively formed 1918-1920
Sources:
Jostein Nordvik (1974). Arbeidarrådsbevegelsen i Norge 1917-19. Hovedfagsoppgave.
Nils Ivar Agøy. (1996). When Officers Need Internal Enemies: Aspects of Civil-Military
Relations in Scandinavia between the World Wars. Journal of Peace Research, 33(4), 469-481.
Retrieved January 7, 2020, from www.jstor.org/stable/424570
Poland
Soldier and worker councils extensively formed 1918-1919
Sources:
https://en.wikipedia.org/wiki/Workers%27_Councils_in_Poland
Romania
Yes, extensive soldier and worker councils formed in Sfatul Țării
Sources:

xxxviii
Andreyev, A., (1971). The Soviets of Workers' and Soldiers' Deputies on the Eve of the October
Revolution, March-October 1917. Progress Publishers.
Sweden
Soldier and worker councils extensively formed 1918-1920
Andræ, C. G. (1998). Revolt eller reform. Sverige inför revolutionerna i Europa 1917–1918.
Stockholm: Carlssons Bokförlag.
Nils Ivar Agøy. (1996). When Officers Need Internal Enemies: Aspects of Civil-Military
Relations in Scandinavia between the World Wars. Journal of Peace Research, 33(4), 469-481.
Jostein Nordvik (1974). Arbeidarrådsbevegelsen i Norge 1917-19. Hovedfagsoppgave.
Switzerland
Starting in 1918, the old guard (Communists) started organizing workers, peasants, and soldiers'
councils. Not known how successful these were.
Sources:
https://rm.coe.int/090000168067fdcd
Yugoslavia
Soldier and sailor councils formed 1917-1921.
Sources:
https://www.jstor.org/stable/pdf/20749812.pdf?refreqid=excelsior%3A656593f7aaed6c269b23
f2c4e431ae67
United Kingdom
Yes, extensive soldier and police strikes together with worker councils (Soviets) 1917-1920
Sources:
White, S. (1974). Soviets in Britain: The Leeds Convention of 1917. International Review of Social
History, 19(2), 165-193.
Wrigley, C. (Ed.). (2002). Challenges of Labour: Central and Western Europe 1917-1920.
Routledge.

xxxix
United States
The first red scare (1919-1920) prompted with formation of soldier and worker councils and a
general strike in Seattle organizing 65 thousand workers between February 6–11, 1919. Seen as a
result of subversive Bolshevism and aimed to subvert American institutions. Incited widespread
repression of worker organizations.
Sources:
https://en.wikipedia.org/wiki/Seattle_General_Strike
The Oregon Socialist Party Bulletin - Workers and Soldiers Council Organized in Portland (1919)
https://en.wikipedia.org/wiki/First_Red_Scare

xl
Appendix B: Revolutionary Fear and the Bolshevik revolution in Norway
1915-1924 – an extended case study
In this appendix, we will provide an extended and thoroughly documented version of the shorter case study
included in the article.
17
This study shows how that various Norwegian elite groups, in the military, secret police, organized business,
and the political establishment, all thought that a worker-led revolution was imminent on at least two points
in time, namely 1918-19 and 1920-21. We also detail that, as a response to this threat, a range of measures
were planned or pursued to avoid the realization of a revolution (Danielsen 1984; Knutsen 1994; Agøy 1994:
73, 136). The more proximate aim of these measures seems to have been to incorporate and strengthen the
reformist part of the labor movement, through various political and economic institutional and policy
changes (Fure 1983, Knutsen 1994, 43-46). To break down our more specific findings, in this case study,
we will document that:
• Labor, as represented by core actors in the Labor Party and several unions, underwent a radical
change following the Bolshevik revolution, strengthening the radical elements at the cost of the
reformists.
• Elites came to believe that revolution was possible and imminent. Furthermore, this change was
directly tied to foreign revolutionary events, Comintern membership and the adoption of radical
ideology in large part of the worker movement.
• Elites responded to their changing beliefs by various measures aimed at “incorporating” the labor
movement politically through electoral reforms and economically through various social policy
elements.
Concerning the economic elites, we mainly consider the Norwegian Employer Association (Norsk
Arbeidsgiver Forening N.A.F.) the leading employer organization in Norway from 1889 (Knutsen 1994).
Regarding the political elites, we consider them as being constituted, at the time, by the residing governments
(Gunnar Knudsen liberal PM 1913-1920 and Otto B. Halvorsen conservative PM 1920-21), conservative
and liberal MPs, and their party organizations.
18
In the state apparatus, we focus on the defense intelligence
office (Generalstabens Efterretningkontor), secret police (Oppdagelsespolitiet), and discussions in the high
command (Generalstaben) and the provisory security commission. The Labor movement is operationalized
as the Labor Party (Det Norske Arbeiderparti, DnA), the Trade Union Federation (Arbeidernes Faglige
Landsorganisasjon, A.F.L.), the youth-wing of the Labor Party (Norges socialdemokratiske
Ungdomsforbund, NU), and the short-lived worker and soldier council movement. We therefore cover all
17
Parts of the material collected for the section on “Proportional electoral system” and “Going revolutionary” is also used in
Anonymized article.
18
Party archives are – unfortunately – missing for our period. The Nazi-occupation in 1940 lead to the confiscation of the archives,
and they were never recovered. Historians have since then used various alternatives sources such as diaries to reconstruct the
preferences and political considerations (e.g, Danielsen 1984). We have corresponded with several leading historians on this period
to ensure that we have consulted most of the relevant material (e.g, Agøy, Brandal, Egge, and Knudsen).

xli
the principle organizations and arenas for coordination and decision making in Norway over the relevant
time period.
Going Revolutionary
The Labor Party (DnA) was founded in 1887, and from the beginning inspired by Marxist thinking. The
first party program stated that DnA had « the liberation of the working classes as our goal, therefore
endeavors the handing over of the means of production to social common property and production changed
from capitalistic to socialistic” (DnA landsmøteprotokol 1891). This line of thinking was then slowly
abandoned as reformism became the leading principle of DnA. Reformism meant a clear adherence to the
importance of the “parliamentary line” in achieving social change (DnA 1918), relying on constitutional
processes of legislation rather than revolution and trying to affect policy-making via extra-constitutional
means. Parliamentarianism was finally adopted by the party program of 1901, following the introduction of
universal suffrage for men, which was adopted in 1898 (Bjørgum 1976, 66-68).
The trade union federation (Arbeidernes Faglige Landsorganisasjon, AFL) was founded on the behest of
the Labor Party in 1899, and was considered a means to achieve changes in wage- and working-conditions
through bargaining with employers. DnA and AFL were bound by the hip, dual membership was practiced,
and having a power base in the unions was a necessary step in gaining the party leadership position. The
unions and party coordinated their demands against employers and politicians, deciding which aims to be
pursued against the employers in bargaining and which to promote in parliament. At the start of 1917, the
reformist leaders such as Olaf Lian held important position in the leadership of DnA and AFL.
Until 1918, the syndicalists had been delegated to a minority position in the DnA. This group mainly
originated from the Fagoposisjonen of 1911 – a syndicalist movement within AFL – and the social
democratic youth organization. Its platform marked a strong break with the dominant “reformist” group.
Fagoposisjonen argued for the use of sabotage, boycott, and sympathy strikes as legitimate weapons against
employers (Olstad 2009, 173).
The leader and primary motor for Fagoposisjonen was Martin Tranmæl. He wrote the group’s ideological
program of 1913, stating that “workers improvements cannot only take place within the established frame
of capitalism, but one must also work to destroy [capitalism] and introduce the socialistic social order”
(Tranmæl 1913, 5). The aim was, to radicalize the union movement, “to make it ready for revolutionary
mass-action” (Bjørgum 2017, 45). While originating in the unions, the movement would interestingly
enough not gain a strong enough foothold to threaten the existing leadership of that organization, aiming
instead to push their interests within the DnA.
The 1917 revolution strengthened the radical movement, especially in the Labor Party. This shift in the
balance of power would develop further, and ultimately lead to a massive change in party-rhetoric, a

xlii
temporary membership in the Comintern, and the expulsion of the reformists social democrats in 1921.
19
Tranmæl’s first reaction to the revolution in March 1917 was to see in it the beginning of “[t]he fall of class-
society and the introduction of socialism” (Bjørgum 2017, 44). For Tranmæl, it was clear that “one can now
clearly see that the fire will spread with a strength such as it cannot be stopped” (ibd, 45). With a reference
to the Zimmerwald program, he demanded that workers organize a general strike against the war, and that
workers arm themselves against the coming counter-revolutionary attempts by the bourgeoisie. Still,
Tranmæl’s attempts at bringing down the reformist leadership of the Labor Party and union congress were
outmaneuvered by the reformist leaders (Bjørgum 2017, 52).
The November revolution would decisively change things in favor for the radicals within DnA (Fure 1983;
Bjørnson 1990; Sundvall 2017). The news from Russia lead to major organizational and ideological
transformations almost immediately. The labor paper, Klassekampen (Class struggle) published the Zimmerwald
declaration to “establish everywhere soldier and worker councils as your body in the struggle for peace!”
(Bjørnson 1990, 509). The movement would establish worker and soldier councils all around the country,
usually outside of the established trade unions (Nordvik 1974). Tranmæl and others would work to set up
and coordinate these councils, including soldier councils. These organizations offered Tranmæl a base of
support that was outside the established frameworks of DnA and AFL, one that the reformists did not
control. These operations also carried risks, as Tranmæl would be charged by the minister of justice for
conspiring to organize military men. Among the resolutions adopted at the worker councils national
conference on March 24 1918 was the immediate introduction of the 8 hour day, and in case the government
ignored their demands, the use of “political mass-strikes” (Bjørnson 1990, 516-517).
By 1918, the radicals had gone from being a minority to gaining enough support to challenge the reformists.
The minority recommendation to the 1918 meeting stated that, “[a]s a revolutionary class party, [DnA]
cannot recognize the affluent classes’ right to economic exploitation and repression of the working class
even if this exploitation and repression is supported by a majority in parliament. The Norwegian workers
party must therefore reserve the right to use revolutionary mass-action in the fight for the economic
liberation of the working-class” (DnA landsmøteprotokol 1918). The position of the revolutionaries, also
those with syndicalist orientation, was to push the party in the direction of extra-parliamentary action, away
from reformism and parliamentarianism. Ole O. Lian, the party vice-chairman in 1918 argued against the
19
It is still debated in the historical community exactly how radical the Norwegian social democrats were, or what was meant by
“revolutionary mass-action”. Sørensen & Brandal (2018) argue that while the social democrats pursued a polarizing and extremist
rhetoric, they were still committed to democracy (see also Bjørnson 1990, 547). In this reading, the Norwegian social democrats
only continued a long-tradition of civil society organization within the bounds of the political regime. We note that our wider
theoretical argument is not fully dependent on whether the social democrats in this period were truly revolutionary, planned to
undertake revolutionary actions, or a full-blown revolution. Instead, our argument is conditioned on the fact that the elites generally
held the conviction - which was strengthened by the international situation and connection - that they were, indeed, revolutionary.
The only historical sources that indicate that this was not the case, (e.g., Hobson & Kristiansen 2001, 169) has been criticized of
not using the available source material (Agøy 2002). Moreover, even these authors argue that the general military command
underwent a decisive change in their understanding of the revolutionary threat in 1917, following news of the Russian Revolution,
and that local agitation now came to be seen as an expression of social radicalism (Hobson & Kristiansen 2001, 135). The seminal
study on the issue, Furre (1983) argues that there is clear evidence of planning for armed rebellion in November 1918, but that these
plans were shelved in late November at the behest of the reformist leadership in the unions.

xliii
belief that a “revolutionary coup” was possible in Norway; the only possible option for influence was to
secure majority support in the population at large.
20
Tranmæl countered that «[t]he majority confuses
parliamentarianism with popular rule. One has a superstitious belief in parliamentarianism. Such a
superstitious belief will lead to a weakening of worker power and to nothing but great disappointments”
(DnA landsmøteprotokol 1918, 41). The reformists lost at the 1918 meeting, and the minority became the
majority. Most importantly, Tranmæl would become party-secretary, with all of the reformists relinquishing
their place in the party-leadership.
In November, in response to the German revolution, the new DnA leadership believed that a revolution
was now becoming possible. Several members of the leadership started working on plans on “arrangements
for a quick takeover of power” between 11 November to 21 November, when the meeting between the
A.F.L. secretariat and DnA leadership ended with the plans being shelved (Fure 1983, 473). Notes from the
planning group show that a coup would be facilitated by the massive organization of worker councils, and
the immediate reforms to be carried out once in power was socialization of the means of production and
the eight hour day. Tranmæl would take the role as “leader of the revolution” (Ibd. 473-476). Did these
measures include the gathering of arms in preparation for revolution? If the reports found in the archives
of the secret police are to be believed, yes. A member among the leadership is reported to have told an
entente agent “weapons and ammunition were assembled on various places (…) When action was to be
taken, workers would be well armed. The weapon stores increased each day.”
21
Another member was tasked
with buying weapons from a German trade boat (Furre 1983, 498). Still, historians estimate that any weapons
that the movement possessed were very limited (Agøy 1994). Even if the main secretariat of A.F.L distanced
themselves from these plans, and this probably was the reason they were put aside (Furre 1983, 498), the
German revolution had strengthened the positon of the syndicalists in the federation. This lead to a
worsening the position of the reformists, which had seen the federation as their bastion against the
radicalism of the DnA party base (Fure 1983, 65).
The radical line would be further strengthened in 1919. At the DnA’s national convention it was declared
that “[t]he party considers mass action in its various forms to be the crucial means in the struggle for
socialism” and even encouraged MPs to strike and soldier councils to be formed and to work against
mobilization (DnA landsmøteprotokol 1919, 12). The vision was a society in which political rights were to
be denied those "that live on the exploitation of others labor. (..) Social democracy will only extend voting
rights to those that work, and connect such a right to work itself." Furthermore, the socialist society must
organizationally be built on worker-, farmer- and fisherman-councils” (DnA Landsmøteprotokoll 1919, 15).
20
Specifically, Lian ruled out any general strikes aimed at the Storting (parliament) since all foods supplies were controlled by the
municipality administration and the state. That meant any action would have to be supported by the A.F.L., which he knew did not
have the necessary resources to support such an undertaking. The only possible outcome from the mass-revolutionary action that
the radicals wanted would be to “knock our own feet from under ourselves” (Bjørnson 1990, 531; DnA landsmøteprotokol 1918,
35-40)
21
(Generalstabens etteretningskontor til sjefen for justisdepartementet 28.11.1918. Den revolusjonære bevegelse i Norge i 1918.)
The seminal historian on DnA in this period (Fure (1983, 479), advises some restraint in interpreting this statement to the agent.
The fellow in question was known to exaggerate, but his statement on efforts being made to secure arms probably reflected some
degree of truth.

xliv
This followed their understanding of the current situation: The capitalistic world was on the brink of
“breakdown (..) Moving in a direction from capitalism to socialism” (DnA Landsmøteprotokoll 1919, 12-
13). Norway was in a “state of maturation for the revolution and socialism”
22
.
The party then also decided to become member of the Comintern. Further, the role of the soldier councils
was expanded from a defensive measure to an offensive strategy to “dismantl[e] the military” (DnA
landsmøteprotokol 1919, 44). DnA also changed its understanding of property rights in the rural sector.
Since 1902, DnA had accepted private ownership in tandem with municipality and state owned lands, and
land-reforms was envisioned to bring about these reforms. With the new land-program of 1919, all land
that was worked by hired help was to be either portioned up to family owners or socialized and managed as
co-operatives. Parliamentary means to policy change were eschewed in favor of a post-revolutionary society
being the only solution to bring about these changes (Fure 1983, 92-93).
The next transformation of the DnA would take place in 1920, with the entry of the party into Comintern.
The vice-chairman Emil Stang had been the single delegate representing the party at the First congress in
1919, and the party would join the Comintern in late 1919. On the second conference, DnA would be one
of the biggest delegations: 10 delegates would arrive in Moscow, including the youth organization leader,
(and later PM) Einar Gerhardsen. It was here that Zinonev, the leader of Comintern, presented what later
was to be nicknamed Lenin’s 21 thesis, outlining the rules for admission to the Comintern.
It is important to recognize the revolutionary nature of the Comintern. The aim of the Comintern, as
recognized by Lenin 21 theses, was to establish organizations that could function as divisions in the
European civil war. The organization was to be divided into isolated cells that could operate independent
of each other, but following a hierarchal system. In other words, the goal was to create communist parties
that could survive the coming world revolution (Sundvall 2017). Adhering to the Comintern theses meant
accepting Lenin’s understanding of the capitalist world as having entered into a state of “civil war”. It also
meant the expulsion of what Zinonev labeled “reformist traitors” in order to avoid the “danger of being
watered down by elements characterized by vacillation and half-measures”.
23
House clearing was therefore
mandatory – even when it came at a loss of important capacities – and so was engaging in illegal activities.
These organizational changes were considered to be needed because “[t]he organization of the party must
be adapted to the conditions and purpose of its activity. The Communist Party should be the vanguard, the
front-line troops of the proletariat, leading in all phases of its revolutionary class struggle and the subsequent
transitional period toward the realization of socialism, the first stage of communist society.»
24
22
DnA landsmøteprotokoll 1919, 12. Social-demokraten. 8 may. 1919.
23
https://www.marxists.org/history/international/comintern/2nd-congress/ch07.htm
24
https://www.marxists.org/history/international/comintern/2nd-congress/ch07.htm

xlv
In 1921, the most ardent of the reformists would break out from DnA. They established the Norwegian
Social Democratic Party, holding on to about 10 percent of the vote. The reformists would not re-join the
party until 1927.
How involved was DnA with the dealings of Comintern? In fact, DnA would take an active role in both
legal and illegal work. Illegal transportation of Soviet propaganda and couriers was carried through by a
secret committee in Kristiania (the pre-1925 name for Oslo), in quite extraordinary proportions (Fure 1983;
468-9). Party offices was used to hide illegals traveling to and from Russia and bribes paid to workers on
the various coast-boats were used to ferry the illegals through customs (ibd. 471). A German spy-report
from this period also outlines that Norway was used as a station to transport funds to practically “the whole
world” (ibd.). Prominent DnA members participated in the smuggling. Einar Gerhardsen after the
Comintern meeting of 1920, for example, would smuggle Emeralds hidden in toothpaste tubes. Another
member of DnA was stopped with 51 gold bars in his luggage.
The actions of the Norwegians were also coordinated with Moscow. For example, a report to the General
Staff’s office for intelligence reveal that there existed a secret agreement between DnA and Comintern,
under which DnA accepted to undertake all orders from Moscow in exchange for political as well as financial
and military resources (Olstad, 1998, 39). Other sources indicate that Norwegian radicals were told to halt
any revolutionary attempts in 1922, in order to wait for more favorable conditions in Sweden which would
facilitate an inter-Scandinavian revolution (Ibd.).
In 1922, the high court barrister and communist Ludvig Meyer is indicated to have been tasked to Research
and detail the possibility of revolutionary action (Olstad 1998, 49). The documents show that he thought
the current year fitting for the beginning of revolutionary action: “Norway is on the verge of breakdown,
which can be exploited by a tax-strike and by pushing for demands that could rally the workers against the
government. The Military and the [kings] guard was not as trustworthy as in 1921, and in the army perhaps,
only as many as 3000 men would go as far as to do their duty” (ibd.). Few public functionaries would go so
far as to risk their lives to fight against the revolution, Meyer considered. The Norwegian middle class was
“weak” and would be easy to contain by the use of terror (ibd. 49-50).
Ironically, the Comintern adherents in the party now pushed for a more moderate line, following new signals
from the third international conference, which indicated the stalling of the world revolution. These DnA
Comintern members would therefore support various reformist proposals. Over time, this created a rift in
the revolutionary wing, which pushed Tranmæl and his ilk to embrace the reformist, social democrats that
had not left the party in 1921. This was arguably also a smart strategic response for Tranmæl. His power-
base was split between the unions and the party, and the ability of the party to gain foothold among the
workers dependent on the unions. He also wanted to pursue a radical line from “below”, using the dual
membership of union and party to build a revolutionary movement. Comintern, with its concept of party
cells, was adamant that dual membership had to be revoked: Against this background, DnA voted to leave

xlvi
Comintern in 1923, the social democrats which had left in 1921 would rejoin the party in 1927. The
Comintern group would leave DnA and establish a Communist party -- “Norges Kommunistiske parti”
(NKP) -- on the 4 of November 1923, with 13 MPs defecting. In 1924, only six of these NKP MPs were
re-elected.
Elites and revolutionary threat
Having described the strong radical turn in both the AFL and DnA, we now turn to how state, business and
political elites over time came to view the revolutionary threat and what responses they considered to be
the best ones. Before we detail these perceptions and responses, Figure VI provides just one illustration of
how perceptions and awareness of the threat-situation, overall, changed during this period. The figure
reports the annual frequency with which the keywords Revolution, Bolshevism, General-Strike and Strikes
were used in articles in four major newspapers (Aftenposten, Dagbladet, Morgenbladet and Bergens Tidene)
-- from the two largest cities, Oslo and Bergen – that were read by various liberal and conservative elites. In
order to facilitate comparison we have included the word “War”.
Figure VI Number of Articles using words linked to fear of revolution compared to that of War in elite
newspapers (Aftenposten, Dagbladet, Morgenbladet and Bergens Tidene)
The revolution of 1917 and the following power shift in DnA fundamentally changed the elites’
understanding of the labor movement. This change took place in both the military, business community
and among the liberals and conservatives.

xlvii
Agøy (1994, 32-34) documents that prior to 1917, both the military establishment and the political elites
shared the opinion that military engagement into internal affairs was to be avoided. By 1918, the elites’ focus
had shifted from external threats to the internal one. Specifically, the general staff had undergone a shift in
their fundamental understanding of labor, and what means were appropriate to use as response. First off,
major steps were taken to set-up “risk-free” military divisions, which could be mobilized during general
strikes or strikes targeting strategic infrastructure. Risk-free here meant divisions excluding members from
the lower classes (Pettersen 2010). The secret police also decided to continue (which would become illegal
after 1918) postage and telegram surveillance of the most radical elements. Especially the clique around the
leader for the radical wing of the labor party, Martin Tranmæl, was put under surveillance. This later led to
Tranmæl being charged for working against the state.
At various points in time, military divisions and battleships would also be mobilized to work pre-emptive
measures against strikes getting out of hand. For example, the battleship “Harald Hårfagre” would be sent
to Trondheimsfjorden during the work-stoppage conflicts in 1918. DnA had put forward the demand and
threatened with work stoppage on the 1 of May unless their demands were met. The commander was told
that «one must be prepared to face disturbances if the decision goes against the wishes of the workers”
(Pettersen 2010, 52). Similar preparations were in place during the railway strike of 1920.
The political elite was even more concerned than the military. As argued by the seminal historian on the
Conservative Party in this period, the labor movement would now come to be viewed as the greatest danger
to the regime since the Thrane movement more than 70 years earlier (Danielsen 1984, 14-15). At the start
of 1918, leading members of the cabinet feared an outright coup (Agøy 1994; 94). The Liberal Party Prime
Minister Gunnar Knudsen therefore established a security commission with a wide mandate to “secure
peace and order if civilian government was brought down” and work to establish guideline for co-operation
between the different defense arms against the syndicalists (Pettersen 2010, 43). The commission would
operate in secret and was meant as a permanent addition, as long as the threat persisted.
Given these preparations, when Knudsen was contacted by N.A.F in the mid of April 1918, which feared
the threat of mass-revolutionary act in the declaration from the worker councils concerning the eight-hour
day, Knudsen’s response was comforting to N.A.F. The military was prepared to intervene if the worst
should come to the worst (Bjørnson 1990, 549). Knudsen would also address the situation publicly in
parliament on March 3: “It is tragic to have to say this, and even more tragic if one ends up having to do it;
but the government is prepared to, that it might become necessary, and if it is necessary, it is better to strike
fast than to show weakness and forbearance opposite possible breach of order, for then it might become so
much worse later on”.
25
The security commission was summoned to convene again on the 12 of November, following the
revolutionary events in Germany. These events had convinced both the political establishment and
25
Stortingstidene 1918 399 (Gunnar Knudsen)

xlviii
especially the military that a revolutionary attempt was imminent also in Norway (Agøy 1997, 75).
Information problems were prevalent, and a primary point of discussion was how keep tabs of the syndicalist
movement (Danielsen 1984, 16). The military secret services and the police were tasked to increase their
surveillance of syndicalist elements, establish networks of informants, and maintain linkages with other
Nordic intelligence agencies. Security talks (and later preparations) would especially focus on the role of
strikers, who were assumed to aim at effectively restricting the government ability to control the country.
Weapons stored in all army deposits around the country were to be made useless, and the preparation of
specially selected army units was initiated. Specially selected here meant the exclusion of soldiers which
potentially could have lower class sympathies (Agøy 1997, 139). These divisions were kept secret for all but
a select few. The literature is uncertain as to the size of these units; Agøy (1997, 118-9) estimate their strength
at around 8 500 troops, supported by additional 4500. Their equipment included machine guns (stressed by
the defense department as needed to ensure crowd control) and field artillery. In a Norwegian context, these
were sizable numbers, a clear indication that the units were intended for civil war. PM Halvorsen
(conservatives) illustrates just how likely the government came to see a revolutionary attempt in a speech to
his fellow conservative MPs before the national rail strike in 1920:
“One is expecting the hardest of civil wars (...) Our present enemy, even with their minority position, would
still be able to win in the moment [and] we cannot know whether they intend to secure the persons of
government. Edvard [Hagerup Bull] therefore said we must secure a government for the nation. He
proposed that Ivar Lykke and Gunnar Knudsen should stand by with their people if anything were to set
the current government out of play. If they and their people were to meet the same fate, the director general
of the finance department should stand ready.” (Quoted in Danielsen 1984, 18).
This quote strongly supports our contention that revolutionary fear was indeed present and credible among
the elites: The quote reveals that the PM was, indeed, setting up the line of succession to a competitor party
and then to the bureaucracy because he believed that the very existence of the regime was threatened.
It is important to recognize that various elites coordinated their responses. For example, the leading
employer organization, N.A.F., coordinated both with the government (under the Liberal politician Gunnar
Knudsen, himself a factory owner), which, in turn, coordinated with the security services (military and police
arms). Our source material shows that revolution was seen as a possibility throughout 1918-1923, by all
these elite groups, even if the various actors believed revolution to be more imminent in some years than
others. For example, N.A.F seems to have been more convinced that a coup or mass-action would take
place in 1918-1919, while the government was relatively more concerned in 1918 and 1920 (Agøy 1994).
First among the legislative responses was a proposal for a law on maintenance of “Public order during war
and under the danger of war“
26
. It was put before parliament 9 march 1918. It gave wide powers to the
executive to act as if in wartime, when war was a possibility, and to define specific areas as zones of conflict,
26
indst. O. XIV 1918, 399

xlix
in which civil authorities could summon the military to maintain law and order. Additional surveillance
measures were also proposed, along the lines of what existed under WWI.
To what extent were these policy responses viewed as radical? Among the conservative (Høire) and liberal
(Venstre) members of the parliamentary committee charged with the treatment, none had any major
objections. The minority position held a somewhat different view. The representative of DnA (Aslaksrud)
viewed the law as such a severe intervention in citizens’ civil rights that it ought to be withdrawn. Yet, also
Johan Castberg (Liberal MP) argued that it broke with the constitution. The parliamentary debates became
quite heated (st. forh. 198 8d s1217-18), with the minister of justice claiming the law was never meant to be
used in peace-time, and Castberg responding that if that was the case, the law was not even necessary. The
treatment of the law was ultimately postponed, and it would never be put before the parliament again.
Several additional military and police measures would be organized against the worker movement. Still our
focus in this case study is on the integrative measures, the silk glove and not the steel hand hidden within.
The silk hand responses were a combination of political and economic reforms, encompassing the
regulations for hours of work, a new electoral system, subsidies for housing, worker councils, profit sharing,
arbitration regulation, socialization of industries, generous unemployment subsidies to unions, and old-age
pensions. We discuss several of these policy initiatives in turn, starting with regulations of work hours.
Eight-hour day and 48 hour week
Before we deal with the politics of the Eight-Hour day reform of 1919 we want to highlight just how radical
a break this piece of legislation was with previous attempts. Before WWI, hours of work were still
unregulated for adults. The only restriction in force applied to children. In 1914, with war raging in Europe,
the “Great Reformator” of the Liberal Party, Johan Castberg, made his second attempt (the first one in 1909
had been a failure on all points) to regulate working hours for adults.
27
At his behest, a proposition was put
before the Storting in 1914, only to be postponed. The proposition outlined two proposals, Castberg (and
the majority of the commission) proposed a 9-hour working day with compensation for overtime. The
minority position, instead suggested a 10-hour normal workday. In 1915, an attempt was again made to have
the act treated in parliament, but the extensive changes it underwent in the Lagting meant that the law ended
up a major disappointment for its original architect. Gone was overtime compensation, daily hours were set
to 10 (not 9), and implementation was set to 1920 with major industries excluded. These changes prompted
Castberg to think aloud in parliament that one might as well drop the whole legislation.
28
Also the employers were initially negative to the proposed regulation: N.A.F even argued that the original
the act of 1909 was already too encompassing, and this was a law that had ended up not regulating hours at
27
Proposals to set hours to 48, 56, 58 and even 72 for men all fell in the Lagting and the Odelsting (Castberg 1914, 2).
28
1915 stortingstidene efterm 20 august page 2368 (Castberg)

l
all. NAF was fully against any regulation of working hours in factories.
29
Increasing demand by workers and
others for such legislation was in itself not enough to push N.A.F. to support working time regulation.
N.A.F was hesitant to resistant on hour reductions also three years later (1918). At this time, bargaining was
highly decentralized, except for in some areas of manufacturing, and N.A.F as a whole did not participate
in bargaining. In one of the first examples of federations becoming involved in localized bargaining, at the
Hydro plant at Rjukan, the N.A.F. foreman had accepted a 51 hours week (down from 54) and resisted 48
hours. When the director placed the proposal for the board, it was voted down. Factory owner Campbell-
Anderson formulated the opposing position by highlighting that it broke with the federations core
principles:
“If I have a principle, and if I mean that this principle has worth, then I must decline: No I do not go into
any form of compromise (..) What we achieve in this way is that it is not us who are changing the ruling
principles”.
30
Giving in on the question of working time regulation was thus still perceived by many companies as to big
a pill to swallow, even after the Bolshevik revolution of 1917. The statement of the organization on 27 of
May 1918 on the proposed extraordinary law eight-hour law illustrates this perfectly: It was not necessary
given current conditions, it would likely to worsen productivity, work in favor of agriculture at the cost of
industry, increased prices on consumer goods, hurt competitiveness and “cause so many so many difficulties,
that industrial stagnation or decline must be expected” (Petersen 1950, 366-367). An eight-hour law was not
recommended by N.A.F. But, at the same time, it would now not necessarily work against the
implementation of such an act. This was a major shift from an organization that only 24 months earlier had
balked at any form of regulation working time.
After July 1918, the situation changed further. Following revolutionary events in Finland and Germany, and
the shift within the unions towards syndicalism, N.A.F would come to accept working hour regulations
both by legislation and in the collective agreements. It was especially the CEO of N.A.F., Lars Rasmussen,
who argued for the necessity of meeting the new ideological orientation of the workers by other means than
force (Knutsen 1994, 29-31). In his new-year speech of 1919 to the board, Rasmussen outlined the situation,
the dangers facing the organization, and the possible way to safety:
“Previously, our organization would respond to such demands with all the means at our disposal. But here
I believe, that we must consider, that behind these demands stand so to speak all the unrest, that in our time
reigns around us on this earth, and it infects also our situation. And the result is that demands become more
sharper, that they become qualitatively different, than they were before. They are born from, if I may be
allowed, by a new concept among the workers, that is, that the time has come for the workers, and now
29
Castberg 1914, bilag 4 page 18)
30
NAF page 42 Sentralstyre 26.06.1918. (Campbell-Anderson)

li
they want to use it. Such as concept must, as far as I can understand, be allowed to unfold, at least to some
degree. Because if we constrain this concept too much, then the pressure might become too great. Then the
development will go on without negotiations, and the result will be that workers say: let us now seize the
day, let us take power. And then we would be stuck in a societal upheaval, a situation that we would, by all
means, seek to avert. We must be aware our times. We have to see its signs and learn its demands. We must
therefore renegade on some of our old principles. We have to embark on new paths and formulate new
ideas. We must act as wise skippers do, not to turn the rudder against the wind and to pressure up against
the coming storm, with the possibility that we may perish. No, we must turn the rudder. We must see to
that we can save what we can save.”
31
It is important to notice how the N.A.F. leader, Rasmussen, reveals the inherent preference pertaining to
the eight-hour day, noting that “before we would have resisted” and that such demands would have been
met with resistance using all available resources. This also fits with previous statements by N.A.F.
representatives. Rasmussen’s quote also support our contention as to the reason for N.A.F. to now support
the eight-hour day: integration of labor into the existing economic system in order to avoid revolution.
Rasmussen was backed by the previous hesitant voices and even by the outright critics such as Campbell-
Andersen. The latter now argued that “I agree that we must look at the signs of our time and therefore there
is no use in resisting the eight hour workday anymore”
32
. One hold-out critic on the board of N.A.F. was
still present, namely Mr. Kopperud. Kopperud argued “that one cannot overcome difficulties by simply
giving up”.
33
He renegaded on this position in May 1919, however, with the words “it no longer lies in our
hands [to act], it lies with the workers, and they will not be stopped”.
34
It is important to highlight that all
statements that we refer to by N.A.F. members here originate from board notes. Importantly, these were
notes that were kept secret and were meant to be destroyed later on (Knutsen 1994). These quotes are
therefore unlikely to represent strategic overtures to unions or meant only as empty rhetoric.
By 1918-1919 various elites had shifted 360 degrees on the question of working hour regulation. To illustrate
the change, we remind that both the Liberals and Conservatives had voted down Castberg’s 9 hour working
day (51 hours per week) proposal as well as the DnA proposal of eight hours per day in 1915. In 1919, all
parties would embrace the eight-hour day as a long lost brother. The social minister Berg (Liberals) led the
charge when he stated that “the times demand social reforms, demand it with necessity (…) we have great
demands for social reforms, and the greatest task in my opinion, is that labor now takes precedence in our
country. (…) Capital should be a servant and helper for labor, but not its master. It is this which is the
demand of our time”.
35
Gunnar Knudsen (Liberal PM) had already by late 1918 decided that the eight-hour
reform was to be made into law and implemented it in his own factories. Knudsen also made it clear to the
31
NAF page 4 Sentralstyre 12.01.1919. (Rasmussen)
32
NAF page 28 Sentralstyre 31.3.1919 (Campbell-Andersen)
33
NAF page 48 Sentralstyre 31.3.1919 (Kopperud)
34
NAF page 61 Sentralstyre 12.5.1919 (Kopperud)
35
1919 stortingstidene efterm 26 march page 646 (Statsråd Berg)

lii
leader of N.A.F. on 30 of November 1918 that the eight-hour day was coming, and that it would be made
into law “without any reservations” (Knutsen 1994, 20). The Liberal government -- here represented by
social minister Berg -- was not shy in detailing the underlying reason for the reform:
“It is imperative, for the sake of our society, that what can be done is done, to ensure that the workers can
feel satisfied in their work, (...) it is with the future in mind of outmost importance for the satisfaction of
the whole of society, that the desire for work is increased. And the department has faith in, that an
appeasement of the old worker demand, of which we are her concerned, to a high degree would contribute
to this. (...) With the eight-hour day implemented by law our country’s workers will find new faith in the
belief that through a development of society as it now is, we can reach a societal-order, where also they may
find their place".
36
We see a clear link between the demands of the working classes and the need for social stability as a key
justification for the legislation. Ensure the legitimacy of the current system also among workers was viewed
as imminent. Both the liberals and conservatives agreed on this point.
37
When on 14 of June and 2 of July
1919 the eight-hour act was put forward to parliament, it passed both in the second and first chamber of
the Storting (Odelsting and Lagting, respectively) by acclamation. Klingenberg (Conservative) stated that “we
will now approve with law a demand that workers in the whole world for a lifetime has declared to be one
of the most important to (...) achieve the social conditions under which they want to live and have a demand
to live under”
38
. Olsen (Conservative) stated in the Lagting that, “I find that one must salute, with both
happiness and satisfaction, that one has come to agreement on such a great issue as this”
39
. This represented
quite a turnaround from parties that had only a few years before strongly resisted the eight-hour day. This
sudden change of heart did not go unnoticed by the socialists. Nygaardsvold (DnA) would lament that
suddenly all parties across the ideological spectrum had come to embrace what they had so vehemently
fought against just 4 years ago:
“The road to legislative reform has been hard to travel. Each time the demand of the workers for an eight-
hour normal-working day was brought forward to the Storting was the demand voted down, or the reform
was so distorted that it would have no impact of consequence for the workers. (...) Workers therefore had
to take on the issue themselves (...). I want to add, that there is no single issue that has to such an extent,
made workers lose their faith in the parliamentary line, that parliamentary action work. (...) As long as
workers did not put any major force behind their demand for to so important demand, the Storting down
voted all demands to reduce working hours to eight hours”.
40
36
Ot.prop. nr 21 1919. pages 8-9 (Berg).
37
Several issues did come under debate and to a vote both in the first and second chamber, such as how to treat firms with
continuous production (Lian wanted 42 hours instead of 48), inclusion of commercial workers, inclusion of rural workers
(landarbeidere), and the distinction in firm size on the application of the law (cut off at 5 employees).
38
1919 stortingstidene 19 june page 141 (Klingenberg)
39
1919 stortingstidene efterm 2 July page 66 (Olsen)
40
1919 stortingstidene efterm 2 July page 64 (Nygaardsvold)

liii
It is interesting to note that Nygaardsvold made clear that the current situation and the commitment of the
working class was connected to the lack of social reforms, with the delay of the eight-hour day being
considered especially egregious.
The eight-hour act for industrial workers followed on the heels of similar legislative enactments for workers
in important state firms and services, namely workers in military owned factories (27 of april 1918), workers
in railways workshops (3 august 1918), and workers on state railways and defense structures(1 july 1918).
Conducive to these reforms were considerations of limiting grievances of workers in strategic state industries
in order to limit the impact of potential mass-strikes. Especially the railways were seen vulnerable to targeted
strikes by the Knudsen Government, and these policies came about after a change in the governments’
perception of a true revolutionary threat (Agøy 1997).
Yet, we should note that revolutionary fear was likely not the only conducive factor to legislation such as the
eight-hour act. To some degree, it had become much easier for proponents to promote a working hour
reform, knowing that the International Labor Organization and the Versailles signatories all had committed
to the eight-hour day. Arguments on maintaining similar rules as the international manufacturing
competition were therefore less relevant than before, and it was clear that what was considered the minimum
standards in labor regulation was changing (Rasmussen 2019; see also statement by Klingenerg and Castberg
in parliament references above). In addition, changes “on the ground”, in the form of collective agreements,
made the elites aware that the initiative was slipping away from the national elites. If they wanted to set the
limits of the new regulation, they had to act now.
Collective bargaining and arbitration awards
Appeasement was not only pursued through legislative changes. Business could also meet (and had to
respond to) worker demands in the collective bargaining system. The 1915 agreements, which had lasted
for over 4 years, were to be re-negotiated in 1919 and the lack of any wage-growth in a period of massive
price increases (no-indexation rules existed in agreements at this time) had led to a decisive fall in real wages.
The collective bargaining results of 1919 were therefore quite generous; N.A.F would accept the eight-hour
day, extensive wage-increases, and the number of paid leave was extended (from 4 to 6 days) in most
agreements (A.F.L. beretning 1919/1920). In the words of Danielsen (1984, 20) “the concessions that were
then granted were greater than in any other single year and, in most employers' opinion, exceeded what the
economy could carry”.
41
In the spring of 1919, The N.A.F Board had decided that one had to avoid ending
up in a major conflict (“avoid setting the score”).
42
Concessions had to be made. Rasmussen argued that
these concessions were to be seen as a form of “insurance” that was necessary to ensure that no “social
upheavals took place, which could lead to things, of which we had no overview”.
43
N.A.F. would later
41
Danielsen uses the word “realøkonomisk”.
42
NAF page 7 Sentralstyre 24.4.1919 (Schuman)
43
NAF page 51 Sentralstyre 23.1.1919 (Rasmussen)

liv
summarize the results of the bargainng as having taken place under “the sign of the revolution”
44
. They
were insurance against giving the “Syndicalist or Bolshevik tendencies (...) a rich soil as to grow and take
root”.
45
Still, they were only to be accepted after some negotiation and preferably be decided under
arbitration.
46
It was important not to appear that one could have given even further admissions.
By 1920, the “good days” were over, so to speak, for the unions; Rasmussen had become convinced (quite
in opposition to the Conservatives in government) that the revolutionary threat was declining. “The
revolutionary waves that we feared last year would envelope us, have, in this year, leveled out somewhat”
47
It was time to regain some of the lost ground of 1919. N.A.F. therefore demanded wage-reductions in all
collective agreements up for re-negotiation in 1920.
N.A.F.’s proposals about wage-reductions were perceived as nothing but a provocation by the unions.
Negotiations stalled. Fearing the increased strife would escalate and spread; the government’s answer was
to use the Temporary Arbitration Act (Fure 1983, 578; Knutsen 1994, 35). In so doing, the government
pushed the fate of the negotiations into the hands of the worker friendly chief justice of the Supreme Court,
Karenus Kristofer Thinn. The Thinnian arbitration decisions -- as they would be remembered -- were
excruciating for the employers. They included that the previous six days of paid leave became fourteen, and
with further wage-increases all over the board. Fure (1983, 578) argues that this was the last true
“appeasement policy” pursued by the political establishment during this period. The Thinnian decisions also
had the effect of stopping any major confrontation that could have led to a change in the basic points
included in the collective agreements. The reason was that the union federation, AFL, had started in 1919
to push for the inclusion of labor into management, and in order to do so, the basic clauses of bargaining
agreements had to be changed. The basic clause laid out employer’s prerogative to manage the firm, and the
“whole history of the employer federation rest[ed] on this clause” (Knutsen 1994, 32).
The workers’ gains would be short-lived, however, and the 1922 bargaining rounds would be nothing like
that of 1921. Gone was Thinn, and as the unions launched what they believed would be the great showdown
with the employers, one that would be so decisive that it would force the government to undertake
fundamental reforms, they found that they had overestimated their strength. Pervasive use of strikebreakers
and the absence of a general mobilization among the populace meant that the “Great Strike of 1921” ended
in a decisive defeat (Bjørgum 1985, 92). Close to all agreements that were up for re-negotiation ended up
with major wage-decreases and the number of vacation days were reduced to eight (A.F.L. Beretning 1922-
23). It was clear that the organizational strength of the unions was on decline in 1922 (Knutsen 1994).
44
Arbeidsgiveren 1. 1920, 3. Oslo.
45
NAF page 22 17 Sentralstyre 16.2.1920 (Rasmussen)
46
NAF page 28 Sentralstyre 31.3.1919 (Campbell-Andersen)
47
NAF page 7 Sentralstyre 20.1.1920 (Rasmussen)

lv
Proportional electoral system
One of the central demands of the labor movement, and especially its reformist line, was electoral reform.
More precisely, the movement wanted the introduction of a system of proportional electoral system (DnA
1918). Over time, the increasing mismatch between the percentage of seats and the percentage of votes for
DnA had increased. This discrepancy was especially driven by the two-rounds voting system of 1905, which
allowed the bourgeoisie parties to coordinate on the candidate best placed to defeat DnA candidates in the
second round.
Danielsen (1984, 19), in the seminal work on the conservative party in this period, writes that “the electoral
reform can stand as a key example on the intention of the conciliatory strategy”. Danielson underlined this
by specifying that the party-leaders in the Liberal and Conservative parties in preparation to the vote had
decided on introducing PR as a way to avoid revolution.
48
The constitutional amendment had been put
forward to the Storting on 12 of December 1917, and would therefore be treated after the next election
(1918). At the start of 1919, it was decided to postpone the treatment of the election commission’s proposals
to the end of the year. DnA would therefore threaten the bourgeoisie parties to bring the proposals to
parliament. Speaking of withdrawing socialist MPs from parliament, or carrying out an “election-strike” the
DnA made a badly veiled threat: “Closing the parliamentary option, will necessarily make the political and
struggles ever more sharp, and the responsibility for this is left with the rulers”.
49
The parliamentary debate on the 28-29 of November shows support for the notion that PR reform was
introduced as a concession to ensure the integration of workers in the current system (thus mitigating
revolutionary threats). The party-leaders had already bargained between themselves, deciding on
implementing some form of PR, even if they were highly split on the specific form of PR to be introduced.
Michelet (Conservative and leader of the Electoral Commission) argued that “[t]he right of suffrage is
introduced in such a way as to breach with the idea of universal suffrage. (...) Most equal influence for all,
rich or poor (..) The electoral system is such that a voice gets a purely different value depending on whether
he lives here or there, and the results have -- as we all know -- become such. This has led to one of the
major political parties, election after election, being under-represented. [This is] an under-representation
that (...) appears to be an institutional consequence of the electoral system itself. We in the Commission are
in agreement to (...) propose proportional elections in some form (..).
50
The DnA representative in the Electoral Commission Magnus Nilssen drew the big picture:
“In 1918 we witnessed that strong forces would make themselves known within the working movement.
(...) We who have participated in the electoral commission were under the strong impression (…) that time
was of the essence in getting a grand proposal ready (…). At meetings and in the press, people from all
48
Several liberal and conservative delegates would complain that the proposal had only been discussed by the various party-
leaders (party presidents), not including the delegates themselves.
49
Dokument nr 24 (1919) fra den forsterkede konstitusjonskomite. Angående valgordningen.
50
50
Stortingstidene 2848 (Michelet). Forhandlinger I stortinget (nr. 366) 1919. Efterm 27 novbr. – valgordningen.

lvi
parties presented the injustice and risk of continuing an electoral system under which especially the Labor
Party was so unjustly treated (…) I am not in doubt, that if parliament was to enact a postponement, it
would do itself a disservice (…) I am highly confident that there will be raised a strong, public and justified
sentiment against parliament, if it does not take to its senses and vote on the current proposition. We all
know how the situation has been in nearby countries, and we have seen the waves wash up on our shores,
and if we are serious in governing our country by legislation and by the parliamentary ways, our duty
demands that we, as quick as possible, arrange the parliamentary institutions in such a way as the people can
be satisfied with it.”
51
Also the leader of the Liberal Party and Prime Minister, Gunnar Knudsen would, albeit in more careful
terms, draw similar connections between the nature of the electoral system and the order of society in his
reply to a DnA MP:
“It was shown that the party he belonged to didn’t receive even half of the representative that it should have
received in proportion to its size of the electorate. This facilitated, (…) within all parties recognized that
circumstances have become such, that this could no longer go on and that it was necessary to establish a
better electoral system, a more just electoral system (…) instead of 41 representatives, that party received
18. It is nothing to be startled over, that such a result can create unwillingness and indignation and awaken
thoughts, which do not go together with a well ordered society.”
52
The leader of the Conservative Party, and later Prime Minister (Halvorsen), would argue along similar lines:
“we demand of all parties that they follow the parliamentary line. It is consequently decisive that we then
open for equal access to all parties to pursue their interests within parliament. (..) The times demands it”.
53
Similar considerations are also found in the statements of individual conservative and liberal MPs. For
example O. C. Müller (Conservative) argued that he was prepared to vote in favor of a proportional system
so “that I would give socialists the opportunity to forward their ideas within the parliamentary lines (...) In
order to secure the peace, as we all so much want and understand is necessary”.
54
Jahren (Conservative) argued that any postponement or voting down of proposals had to be avoided,
especially because “that by my vote I might hamper a large party in our country [DnA] to work as it has
been said following parliamentary lines. I believe that one must contribute what one can within reasonable
limits to support that direction that wants to work to the betterment of society within the parliamentary
lines”.
55
Christain Rud (Liberals) proposed that the current electoral system was “without principles and
arbitrary” and further noted that “[t]he demand of the social democrats for fair representation in parliament
51
Stortingstidene 2878-9 (Nilssen). Forhandlinger I stortinget (nr. 360) 1919. Efterm 27 novbr. – valgordningen.
52
Stortingstidene 2930 (O.C. Müller). Forhandlinger I stortinget (nr. 366) 1919. Efterm 28 novbr. – valgordningen.
53
Stortingstidene 2876 (Halvorsen/president). Forhandlinger i stortinget (nr. 360) 1919. Efterm 27 novbr. – valgordningen
54
Stortingstidene 2921 (O.C. Müller). Forhandlinger I stortinget (nr. 366) 1919. Efterm 28 novbr. – valgordningen.
55
Stortingstidene 2929 (Jahr). Forhandlinger I stortinget (nr. 367) 1919. Efterm 28 novbr. – valgordningen.

lvii
must at most opportune moment be accommodated: no unnecessary postponement must take place”.
56
Minister of justice, Blehr (Liberals), noted that “it was stated by the socialists first representative, that one
is kicking the chair under his party’s parliamentarianism, if one does not solve this problem [of electoral
injustice]. Yes, I do not think parliament should do so. Instead, one should endeavor to secure that
development can proceed in a calm way (...). Some things might feel as a sacrifice, but one should act so in
that feeling, I mean, that this sacrifice, I bring happily when it turns out that it is necessary for a good and
healthy development for our country”.
57
Socialization of property and worker participation in management
Prior to 1918, the questions of the involvement of workers in firm management or socialization of property
were anathema among Norwegian political and economic elites.
58
As already outlined, the position of the
elites was quite clear that any infractions on the right of firm owners to manage their company was
unacceptable (Petersen 1950, 337). With the Bolshevik revolution of November 1917 and the adoption of
a radical line within DnA in 1918, the debate on the right of workers to manage change radically. The
Norwegian worker movement, spearheaded by Tranmæl, advocated the formation of company councils
(“Bedriftsråd”) within an overarching frame of branch and national councils. These were conceptualized as
decision-centers: management of firms and whole industries were to be shifted into the councils as a first
step to socialization (Petersen 1950, 337-8). This was a question that broke with N.A.F.’s “fundamental
principles” and for which the organization was ready to mobilize “the whole strength of the federation”
(ibid.).
While the business elites did not intend to play ball, the political elites went as far as taking the initiative in
pursuing the question. In the face of the changing situation of 1918, the Liberal government proposed to
deal with the increasing social conflict surrounding profits by firms in the traded sector with distributing
employee dividends. This new position was launched in the Opening of parliament speech (“Trontalen”) in
1918. Astoudninly, the issue was even supported by the leader of the Conservative group in parliament:
“What is unfortunate is (…) that the traded interests have taken out exorbitant dividends. Furthermore, it
is unappealing that the traded interests also under these favorable conditions have held their employees’
wages down at such a level (...) it is completely unjustified. (...) The state cannot stand indifferent in the face
of these hardships. For these reasons, I find it regrettable that no employer has even in such a restricted
extent experimented with having their employees revising dividends or share in the companies themselves.
(...) I now view with satisfaction that the government party now wants to address this issue”.
59
56
Stortingstidene 2955-6 (Rud). Forhandlinger I stortinget (nr. 368) 1919. Efterm 28 novbr. – valgordningen.
57
Stortingstidene 2893-4 (Rud). Forhandlinger I stortinget (nr. 362) 1919. Efterm 28 novbr. – valgordningen.
58
st.prp. nr 134, page 1. 1918 – Bevilgning til Arbeiderkommisjonen.
59
Stortingstidene 534 (Bull). Forhandlinger I stortinget (nr. 67) 1918. Efterm 16 mars. – trontaledebatten.

lviii
He was supported by Peterson (Conservatives, factory owner), who linked dividends shares explicitly to the
possibility of lessening current worker troubles:
"I would say that if there exists antidotes that are congruent with our view of capitalist society, which can
also bring in an additional element of greater social peace, greater trust, and greater harmony interests
between capital and labor, between the factory and its workers, then it is the greatest task which lies before
us. Many have argued that we have the solution in the dividend system. (...) The government should put
down a commission to evaluate this question."
60
The result was, in fact, that the government set down a commission to evaluate “the question of employer
dividends and share of management”. The aim was social preservation; “to create a closer community of
interest between industrial firms and their workers by making these partake in the company’s capital and
dividend as well as their management”.
61
It would later be named the Worker Commission of 1918 (WC),
and ended up putting forward legislative proposals on work councils (1919, 1920) and profit-sharing
schemes (1922). The only real impact of these efforts would, however, be the temporary act on work
councils of 1920.
Even more radical was the Socialization Commission (SC) of 1919. Its task was to evaluate not only the
question of socialization, in general, but also socialization of specific firms and industries such as “Norsk
Hydro” “Meråker Bruk” “Odda smelteverk”, and the complete mining industry (Danielsen 1984, 19). The
1919 commission was wanted by DnA, but initiated after a demand from the workers at the “Hydro Plant”
in Rjukan. Several committee members of the WC also participated in the SC, meaning that it would not
hand in its report until after the work of the WC was completed.
In addition, a Land Commission of 1919 was set-up to discuss the issue of inequality in arable land. Its
primary task was to assess the possibility of procuring land for tenants. Yet, little came out of its efforts.
On the issue of employee participation in management, the WC decided in early 1919 to develop an
anticipatory proposal. The WC quickly split into a majority consisting of worker representatives and
university academics, and a minority consisting of business leaders. The majority proposed the establishment
of a system of work councils, with firm, district and national councils. Work councils were not only to play
an advisory role. Instead, they would have executive power over questions of introduction of new work
schedules, procurement of technology, and in hiring of supervisors.
62
The minority would rather propose
a system of firm-specific councils, and their power was to be advisory only. It was the minority position that
was put forward to parliament by the Halvorsen (Conservative) government in 1920 – with certain small
modifications – and enacted.
60
Stortingstidene 596-597 (Peterson). Forhandlinger I stortinget (nr. 67) 1918. Efterm 16 mars. – trontaledebatten.
61
st. prp. nr 134 .page 1 1918 – Bevilgning til Arbeiderkommisjonen.
62
ot.prop. nr. 65 om utferdigelse av en midlertidig lov om arbeiderutvalg i industrielle bedrifter

lix
In 1922, the WC outlined both its proposed company council scheme and its proposal on profit sharing.
The group had again ended up being split, with the commission writing two major opposing legislative
proposals. The majority position proposed profit sharing to be introduced in all firms (public and private)
except in the primary industries. Dividends would be paid proportionally in wages and “capital wages”, the
latter being set by each firm. More radically, a fund, to be made up of taxes on profits from firms exceeding
10 percent, was to be set up. This fund would be used to buy up defunct firms and establish new firms. This
proposal was dead on arrival (Petersen 1950, 355).
Even with only the limited proposals winning support in parliament, the business elites were far from
satisfied. N.A.F considered the 1920 Work Council act as “pure madness” as a policy in itself, but as a
defensive measure against syndicalist tendencies and revolutionary threats it made sense.
63
Still, N.A.F. and
employers did not publicly convey this preference. Instead, it was stated that a reform “would create greater
job satisfaction and increase work performance” (Danielsen 1984, 32). N.A.F. leaders were even more
negative of the profit sharing scheme of 1922, but more pertaining to the way the proposal was structured
than profit sharing per se. In 1918, N.A.F. had set-down its own commission to evaluate the question of
profit-sharing (Petersen 1950, 350). Its conclusion in 1921 stated that while voluntary forms could be
implemented satisfactory, “it is unimaginable that a compulsory arrangement can be implemented with
satisfactory results (ibd., 352)”. Still, N.A.F. saved its hardest criticism for the socialization proposals “which
only built on unsatisfactory knowledge of the most simple and elementary social and economic conditions,
and one could only regret that the above mentioned fraction did not have the necessary insight on this
area”
64
The radicals within the labor movement were concerned about the dangers of an elite-initiated reform
surrounding company councils. If the role of the councils were to be too narrow, and only focus on specific
aspects of production, only have the power to suggest and not to vote through changes, their role as
revolutionary institutions would be quite limited (DnA landsmøteprotokol 1920). They would instead take
the form of institutions aimed at “conciliation between the classes” (ibd. 76). In its place, DnA supported
the majority commission’s proposal, which would have given the workers greater influence over company
decisions. Perhaps somewhat surprisingly, the unions advocated against profit sharing. The reason was
considerations as to worker solidarity within industries and support for the centralized collective bargaining
system. The union wage-bargaining strategy was to force wage increases in high-wage firms and sectors (also
likely to have higher profits) and to then use these increases as an argument for pushing up wages in low-
wage firms. Profit sharing would make workers in successful firms less willing to support the union strategy
since they would benefit from the .
How did individual employers respond to the WC and SC commissions? The WC and the SC members
found that employers were far from willing to collaborate. This was especially the case when it came to the
63
NAF page 20 Sentralstyre 23.08.1920 (Rasmussen)
64
Expert from the N.A.F. run newsmagazine “Arbeidsgiveren” 24 of November. Cited in Peterson (1950, 355)

lx
collection of statistics, which lead the commissions to, in 1920, request the Storting to pass a new law
concerning its right to collect data. Lian (DnA) would bring this issue out in the open when he remarked in
parliament that “I just want to say that I have a strong feeling, a feeling that there has been an underground
resistance against this task which the government has entrusted the commission. (...) This is the reason that
the commission has come here to parliament to be granted the protection of the law to gain the necessary
information to carry out its work”.
65
The act would pass parliament, but it did signify the strong opposition
of the employers against even evaluating the thorny question of socialization and profit sharing.
As the fear of revolution subsided significantly with the defeats of the various strikes and fall in
organizational strength, the political elite shifted in the question of socialization. The Conservatives, which
had all voted for the continuation of the SC up until 1922 would in 1923 abruptly change their perception
of the necessity of such a scheme.
Commissions also had to have their yearly budgets approved by the Storting, a process that had in
Norwegian parliamentary history tended to result in acceptance by acclamation. This was, however, not the
case for the SC in the current climate. The leader of SC estimated that one would need an additional 30 000
NOK to carry out the SC’s tasks, but this was then cut down 12 000 NOK by the head of the
“Lønningskomitee” in the Storting, Lykke of the Conservative party. The mandate was further reduced to
only give a theoretical elaboration of the question of socialization, leave out the issue of socialization of
specific Norwegian firms, and not to draft a legislative proposal. Labor party representatives put forward
counter-proposals. The debate became heated, and labor candidates argued that the commission was being
assassinated, pleading Lykke to not “not to start a class-warfare on an evaluation of a question such as
this”.
66
Madsen (DnA) pointed out that the committee foreman was a High Court justice, and that the
12 000 kroners wouldn’t even cover his own yearly wage of 14 000, which de-facto meant that he would
have to stop his work in the commission. In the end, Madsen asked
“Shall one now stop [with the work of the committee]? (...) This is in my opinion a bad use of state money,
that one first sets in motion a great endeavor and so, when it is practically finished, cut it off and say, that
of this we want nothing. It shows Høyres [the Conservative Party] fear of having these matters evaluated;
Høyre is afraid that the capitalistic society does not survive the light of day; it does not survive being
investigated by a public committee”.
67
Lykke’s proposal won the day against the combined vote of the various labor parties. What is the likelihood
that Lykke was acting in good faith? Was he just restraining a committee that had gone overboard, by more
precisely stating its mandate? We find this less likely- First off, Lykke had showed no problem with WC and
SC commissions previously, as he had even been a participant in the WC commission until entering
65
Stortingstidene 573 (Lian). Forhandlinger I stortinget (nr. 72) 1920. Efterm 1 novbr. – sosialiseringskommiteen.
66
Stortingstidene 673 (Gjøstein). Forhandlinger i stortinget (nr. 72) 1923. Efterm 23 March. – kommisjonsbudgsjettet.
67
Stortingstidene 675 (Madsen). Forhandlinger i stortinget (nr. 72) 1923. Efterm 23 March. – kommisjonsbudgsjettet.

lxi
government in 1920. The committee had been granted funding that it had asked for each year up to 1923,
without any complications. It is more likely that, as the revolutionary threat dissipated and the balance of
power tipped in favor of the elites, Lykke and others now felt confident enough to push back against the
appeasement policies of 1918-20.
Such a change is also traceable in the party manifestos of the Conservative Party. In 1909-1915, the
manifestos exclude any mention of any form of co-ownership or co-management. The manifesto of 1918
is therefore unique; here it is stated that the issue of co-ownership must be “evaluated and solved” (Hørie
1918). However, these lines did not make it into the 1924 manifesto. The change in the position of the
conservatives is nothing but startling if viewed as a change in the attempts to formulate a long-term
conservative ideology. But, the change fits very neatly within a perspective of strategic response to
revolutionary pressure. The strategy could be, respond to socialization demands in period of weakness by
setting-down a broad commission with a wide mandate to collect and process data-material, and with the
power to propose a legislative amendment. As the revolutionary period have subsided, slowly push back
against the commission by challenging its mandate and then hamstringing its ability to act.
In sum, the work of the commissions discussed in this section would, in the end, amount to little. The law
proposals all came to parliament at a time in when the revolutionary threat was on the decline, both
domestically and internationally. It is therefore not surprising elites worked hard to ensure that nothing was
to come from the quite radical proposals proposed by these commissions.
Old-age pensions
Among the Scandinavian countries, and even among the rich western-European industrial nations, Norway
was one of the last countries to implement a government-regulated old-age pension. Ironically, Norway was
one of the first countries in the world to propose an old-age pension, in 1844. This was followed by similar
proposals by individual MPs in 1854, 1868, 1869 and finally a more formal proposal on the introduction of
an old-age pension originating in the Worker Commission of 1894. A proposal was finally put forward to
the Storting in 1899, but was then not put to a vote. In 1902, a new commission was established in order to
further evaluate the question, without any new proposals being formulated. In 1907 the People’s Insurance
Committee was charged to evaluate the question. It handed in two proposals that would frame the pension
debate for the near future: one tax-financed proposal favored by DnA and one insurance scheme favored
by the Liberal Party. The question was consequently politicized, all three major parties had pensions (using
different wording corresponding to their preferred solution) as early as 1909. Still, no proposition had been
put forward to a vote in parliament even before WWI.
Coinciding with the revolutionary situation in 1918, the government unveiled the first royal proposition on
an age and disability pension scheme, based on the earnings-related scheme preferred by the liberals in
government. The system would only insure people of working age and the first pensions would be paid out
14 years after the implementation of the scheme. All workers over 56 years old were therefore excluded.

lxii
The proposition was put to treatment in parliament in 1919. The conservatives were highly critical of the
proposed scheme, as it would mean the implementation of an extensive bureaucracy to control and manage
the individual contributions.
68
DnA wanted a more progressive system and pensions also for the elderly at
the time of implementation. Thus, the Storting unanimously decided to evaluate a tax-proposal to go along
with the earnings-related alternative of the liberals. This was the first time parliament had agreed on the
issue of old-age pensions.
The push for a tax-based system was also a major break in the line of the Conservatives. Before 1919, the
conservatives had always preferred an earnings-related system, but now they sided with DnA in pushing for
a tax-based system. As with the introduction of PR, this was an initiative from the party-leadership among
the conservatives (Pettersen 1978, 13). The aim was to create a society-preserving institution, using the olds-
age pension as a basic security without the stigma of poor relief (Danielsen 1984, 30). It was decided to
evaluate the tax-alternative in a social-commission, which handed in its proposal in 1920, but the
government fell before it could put forth a proposal to a vote. In October 1920, the new Liberal government
put forward an Odelsting proposition outlining a new version of the earnings-related scheme. In January
1921, Halvorsen (Conservative) put forward a revised tax-scheme proposal, but the minority of the Social
Committee in parliament wanted an earnings-related alternative instead. The situation ended up in a vote
for or against postponement, and the Liberals won the day with 45 vs 40 votes. Treatment of the two
proposals was attempted during the summer of 1923, but it was again postponed, until October, when it
was again treated in the Odelsting and finally made its way to Lagting.
When reaching the Lagting on the 30 of November 1923, discussion and disagreement was still prevalent.
Liberal MP Høgseth decried the increasing burden of state and municipality budgets, accusing the
proponents of the tax system of communism.
69
The social minister Klingenberg (Conservative) was harsh
in his reply, exclaiming that: “No, this has nothing to do with communism or socialism. If it is anything, it
is policy aimed at the preservation of society!”
70
Most importantly, if no legislation was passed, the
consequences could be dire: “if we go to society and say: no, we may have promised this for more than 30
years, and worked with it 30 years, but you shall not have it.I If so, I think one will see a reaction of the kind
that will bring no joy to anyone except those that wish badly for our society”.
71
Nygaardsvold (DnA) replied,
“I agree that this reform, is preferable to reforms that one has taken in other countries to guard against, that
the existing capitalistic society should be not be overthrown”.
72
The Social Democrat Magnussen
underlined the consequences of postponement: “One talks so often in these times about the working class
and its attitude to the parliamentary system of governance, its position on parliamentarianism. I think that
68
The 1918 proposal outlined contributions in proportion to income and property and that the payment was to take into account
the number of children and the previous contribution record. All of these considerations suggested the need for a growth in the
state and municipality bureaucracy, which, at the time, was seen as highly un-manageable.
69
Stortingstidene 249 (Rud-). Forhandlinger I stortinget (nr. 32) 1923. Efterm 30 novbr. – alderstrygden.
70
Stortingstidene 257 (Klingenberg). Forhandlinger I stortinget (nr. 32) 1923. Efterm 30 novbr. – alderstrygden.
71
Stortingstidene 259 (Klingenberg). Forhandlinger I stortinget (nr. 32) 1923. Efterm 30 novbr. – alderstrygden.
72
Stortingstidene 264 (Nyggardsvold). Forhandlinger I stortinget (nr. 32) 1923. Efterm 30 novbr. – alderstrygden.

lxiii
one should consider the consequences of saying no to a reform such as this. One should not disappoint a
working class, which has put its hope and trust to parliamentarianism and is waiting on what the Lagting
decides. One should think not just once, but two and three times over”.
73
Yet, with the failure of the DnA proposal, the remaining question was simply whether the government was
to commit to a principle for how old-age pensions were to be organized. In the end, the tax-based option
was chosen by the combined efforts of DnA, the many of the Conservatives, and some of the Liberals.
The old-age pensions proposal would mark the end of a long-line of policy proposals initiated by
conservative and liberal politicians to starve the revolutionary beast. At the time, trade union organization
had declined to pre-war levels, the failed great strike of 1922 showed the weakness of the radical line, and
DnA would leave the Comintern in 1923. Internationally the Red Army was defeated in Poland. The failed
revolutionary uprising in Bulgaria in 1923 only reinforced the notion that the “World Revolution” was,
indeed, dead.
It is therefore only fitting, that the 1923 act on pensions was never implemented or even brought before
the Storting to a final vote. It would go the way of the socialization proposals: several legislative initiatives
and proposals, new commissions and new postponements. In the end, only the principles were enacted, but
never implemented. Old-age pensions would have to wait another 13 years. In 1936, Nyggardsvold (DnA)
had become PM for a labor-agrarian majority coalition. He dusted of the old proposal of 1923 and finally
pushed it through parliament to implementation.
The Signals of Revolution
The historical sources clearly suggest that perceptions of revolutionary fear was a key driver of the different
legislative and other policy initiatives, some of which were adopted and others dropped once the
revolutionary threat resided. A primary cause of this domestic revolutionary threat in Norway was the
Bolshevik revolution (Agøy 1994, 74; Sundvall 2017), reinforced bythe revolutionary and civil war events in
Germany and Finland (Fure 1983). In line with our expectations, international organizational linkages were
important to elites in ascertaining the level of revolutionary threat. In reports and discussions, membership
in international organizations were used as indicators of revolutionary sentiment among labor groups. The
defense intelligence office for example writes, already in 29 February 1918, that a “named source could
report that Norwegian participants in the Zimmerwald-conference now was arming themselves in
congruence with the conference decision to carry out armed rebellion” (Agøy 1994, 76-77).
The radical DnA MPs and extra-parliamentary parts of the labor movements were also strategically working
to foster the impression of a coming revolution, of radical syndicalist challenging the existing order. On the
other hand, also reformist leaders could seem to overplay the dangers of the syndicalist, revolutionary threat,
73
Stortingstidene 265 (Magnusson). Forhandlinger I stortinget (nr. 32) 1923. Efterm 30 novbr. – alderstrygden.

lxiv
perhaps in order to build trust with the elites, ensure future policy concessions for their constituents, or
secure their own political position. The best example of a reformist leader doing so, and perhaps being
motivated by all the three mentioned concerns, is perhaps the 20-year long chairman of the A.F.L. and DnA
MP (1916-21) Ole O. Lian and his dealings with the N.A.F chairman Rasmussen between 1918-1922. The
rising syndicalist movement under Tranmæl threatened to usurp Lian’s position. Lian and other reformists
therefore tried to use revolutionary fear among the conservatives in order to achieve the reforms they
thought would convince their own members to support the reformist line. At the same time, Lian held
secret meetings with the N.A.F. chairman, stressing that not all demands were seriously meant. He even
proposed various ways in which the employers would appear to meet the demands of the various syndicalist
unions, but which would never be enforced. These proposals were meet with astonishment among the
N.A.F. board members and were not pursued, as they were considered to possibly backfire and threaten the
position of their friends in the unions (Knutsen 1994, 34-36 40-43).
In the debate preceeding the PR-vote in 1919, we find that reformist DnA MPs in parliament adopted this
tactic. MP Gjøstein didn’t mock about, noting that “there are strong forces in the direction of abandoning
the parliamentary line, because parliamentarianism has not delivered (...) I would go as far as one possibly
can go in maintaining the parliamentary line (...) but if that is to happen we must have an electoral
[reform]”.
74
MP Buen proposed that “it is unjustifiable to continue even one more day with our current
system (…) if we do not get a positive result, this would equal to kick to legs below the parliamentary work.
The consequences I shall not discuss, but the responsibility rests with those that have created this
situation”.
75
MP Foshaug stated that “if one wants to continue down this path, one cannot demand that
the working class that struggles to promote its cause through the parliamentary channel, can continue to do
so (...) I would therefore most empathically warn against postponement”.
76
Again we see a rhetoric of
“inclusion or else” employed by reformist DnA MPs, which can be interpreted both as a direct warning (as
similar language was employed by the radicals), but also as a way to strengthen their position against said
radicals.
Signaling of revolutionary threat was not only a matter of rhetoric but also a function of the wider
international situation, and particularly of DnA becoming member of the Comintern. Agøy (1994, 139) for
example, argues that it was hard for the military, socially and politically distant from the labor movement,
to know how to interpret exactly what was going on in these tumultuous times. They would therefore rely
on easy to identify signals:
“Times were uncertain: Germany was since March rocked by rebellions; in Great Britain far-reaching
emergency laws had been enacted and a society wide paralyzing miner strike had just been averted as late as
April-May 1921. For officers with limited knowledge to the working movement, it could be hard to ascertain
74
Stortingstidene 2859 (Gjøstein). Forhandlinger i stortinget (nr. 359) 1919. Efterm 27 novbr. – valgordningen
75
Stortingstidene 2875 (Buen). Forhandlinger i stortinget (nr. 360) 1919. Efterm 27 novbr. – valgordningen
76
Stortingstidene 2929-30 (Foshaug). Forhandlinger i stortinget (nr. 367) 1919. Efterm 28 novbr. – valgordningen.

lxv
what the quite open and visible struggle within the movement was truly about. What they did notice, was
that known worker-leaders publicly, and with great vigor, was eager to transform the strike in a revolutionary
direction. They, and the rest of the Norwegian bourgeoisie, took notice that DnA on the 27 of march
accepted the Moskva-theses.”
DnA’s membership and role in the Comintern was a clear signal of the revolutionary intent and capacity of
the (for a few years) leading radical wing of the party.
The signaling seem to have worked. Changes in the party programs of the Conservatives provides one
indication of this notion. Comparing the 1909 program to those of 1918 and 1921, the Conservative Party
had gone from being silent on the issues to promising to secure measures against unemployment and lack
of housing, and even to evaluate the question of employees’ dividends.
77
Only old-age pensions had been
present in the 1909 program. Yet, in the 1924 program, none of these issues was any longer mentioned. The
necessity of social policy innovation had dissipated in tandem with the credibility of the revolutionary threat.
Summary of “Appeasement Policies”
the Bolshevik revolution and subsequent formation of Comintern sparked a decisive change both in the
radicalism of the Norwegian labor movement, but also in perception of this movement as a radical and
potential revolutionary force amongst the elites. A combination of repressive and inclusionary tactics was
developed and coordinated by the various political, military and economic elites. The aim was to mitigate
grievances by including labor in the existing economic and political system, but also develop contingency
plans in case grievances would remain high and disorder result. The elites worked together to weaken the
radical groups, and simultaneously strengthen the reformist groups, both in the trade unions and DnA (and
splinter parties). Still, even when faced with what they believed to be revolutionary forces, both political and
economic elites used stalling tactics to avoid adopting measures that would clearly violate their primary
interests, notably in the economic realm. This bet paid off, as the revolutionary threat soon subsided.
77
The conservatives (Høires valgprogram 1918) point number three on their electoral manifest mentions various social issued to
be solved or policies introduced: effective measures for at remedy housing demand (bolig nød). Peoples insurance to be introduced
( Folkeforsikringen gjennemføres, early name for pensions). Question of having employees and wageworkers receive dividends
should be evaluated and solved. In 1921 (Høires valgprogram 1921) their manifesto also included measures against the extraordinary
high unemployment.

lxvi
References not listed in main document used in the appendix
Albertus, M (2015). Autocracy and Redistribution: The Politics of Land Reform. Cambridge University: Cambridge.
A.F.L. (1919/1920). Beretning. Arbeidernes Aktietrykkeri: Oslo.
Agøy, N. I. (2002) Forsvarshistorie på gyngende grunn. Tidsskrift for samfunnsforskning 3: 587-595.
Bjørgum, J. (1977). Industrialisering og radikalisme. Replikk til William M. Lafferty. Tidsskrift for arbeiderbevegelsens
historie (TFAH) 2, 131–142.
Bjørgum, J. 1985. LO og NAF 1899-1940. Tidsskrift for arbeiderbevegelsens historie. 2: 85-114.
Pettersen, P.A. (1978). Alderstrygd som prinsipp eller politikk : om den sosialpolitiske stagnasjon i mellomkrigstiden.
Hovedfagsoppgave. Oslo. UIO.
Høire (various years). Valgprogram. Archived by NSD: Partidokumentarkivet.
https://nsd.no/polsys/data/parti/partidokumentarkivet/
Olstad, F. (2009). Landsorganisasjonens historie: bind 1. Oslo: Pax forlag
indst. O. XIV (1918). Oslo.
Ot.prp. nr 21 (1919). Om utferdigelse av en lov om forandringer i og tillegg til lov om arbeiderbeskyttelse i industrielle virksomheter av
18de september 1915. Oslo
Nordvik, J. (1974). ”Arbeiderrådsbevegelsen 1917-19”, Bergen Hovedfagsoppgave, UiB,
Tranmæl, M. (1913). Hvad fagoppositionen vil. Note, program. Trondheim.
st.prp. nr 134 (1918). Bevilgning til Arbeiderkommisjonen. Oslo.
Rasmussen, M (2016). How organized labor shaped the origin and structure of the regulatory revolution 1870-2000.
Forlaget Politica: Aarhus.
Sørensen, Ø. & N. Brandal (2018). Det norske demokratiet og dets fiender Oslo: Dreyer Forlag.